article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
recently , the ordered double perovskites sr@xmath0@xmath2@xmath3o@xmath1 ( @xmath2 being the magnetic @xmath4 element , and @xmath3 being the @xmath5 or @xmath6 element ) have attracted a great deal of attention .
the interest to these systems was spurred on by large intergrain - tunneling magnetoresistance observed at room temperature in sr@xmath0femoo@xmath1 and sr@xmath0fereo@xmath1 @xcite , which opens wide perspectives for technological applications . on the contrary , sr@xmath0mnmoo@xmath1 remains paramagnetic down to the very low temperature @xcite .
the magnetic susceptibility of sr@xmath0mnmoo@xmath1 obeys the curie - weiss law with fairly large effective moment . however , there is something that prevents formation of the long - range magnetic order in this material ( a similar situation occurs in double - layer manganites la@xmath7sr@xmath8mn@xmath0o@xmath9 around @xmath10@xmath11@xmath12 @xcite ) . in this work
we argue that such a behavior may be due to the competition between ferromagnetic ( fm ) and type - ii antiferromagnetic ( afm ) phases .
the main idea of this work is to complement the standard total energy calculations with the analysis of inter - atomic magnetic interactions in the collinear magnetic state characterized by the directions @xmath13 of the spin magnetic moments .
the latter can be described in terms of a perturbation theory expansion for small rotations @xmath14 near @xmath15 .
the perturbed
direction of the magnetic moment at the site @xmath16 is given by @xmath17@xmath11@xmath18@xmath19@xmath20@xmath21@xmath22$]@xmath23 , and the total energy change @xmath24@xmath11@xmath25@xmath26@xmath27 in the second order of @xmath14 can be exactly mapped onto the heisenberg model as @xcite : @xmath28 .
\label{eqn : heisenberg}\ ] ] eq .
( [ eqn : heisenberg ] ) originates from the taylor series expansion for @xmath29 near @xmath30 , and @xmath31 are related with the second derivatives of @xmath29 with respect to @xmath32 .
@xmath31 can be expressed through the matrix elements of the one - electron green function and the magnetic part of the kohn - sham potential at the site @xmath33 , @xmath34@xmath11@xmath35@xmath26@xmath36 , as @xcite : @xmath37 with @xmath38 denoting the trace over the orbital indices .
all calculations have been performed in the local - spin - density approximation , using the asa - lmto method @xcite .
the calculated densities of states for the fm and afm phases are shown in fig.[fig.dos ] , as a function of the breathing distortion @xmath39@xmath11@xmath40 ( the ratio of mn - o and mn - mo bondlengths ) . when @xmath39@xmath41@xmath42 , the fm state is metallic . the fermi level crosses mn(@xmath43 ) band in the majority ( @xmath44)-spin channel and the mo(@xmath45 ) band in the minority ( @xmath46)-spin channel .
therefore , one can expect two main contributions , which tends to stabilize the fm ordering : the canonical double exchange ( de ) operating in the @xmath44-spin mn(@xmath43 ) band , similar to the colossal magnetoresistive manganites @xcite ; and a generalized de mechanism associated with partial filling of the @xmath46-spin mo(@xmath45 ) band and originating from the strong mn - mo hybridization @xcite .
the breathing distortion @xmath39@xmath47@xmath48 tends to fully populate the mn(@xmath43 ) band and depopulate the mo(@xmath45 ) band .
therefore , when @xmath39 increases , the de interactions of the both types will _
decreases_. the afm phase is metallic in the undistorted cubic structure ( @xmath39@xmath11@xmath48 ) .
however , even small oxygen displacement towards the mo sites opens the band gap .
the situation corresponds to the formal configurations @xmath49 ( mn@xmath50 ) and @xmath51 ( mo@xmath52 ) of the transition - metal sites .
the half - filled mn(@xmath4 ) states give rise to the afm superexchange ( se ) interactions , similar to the rock - salt monoxide mno , which are roughly proportional to @xmath26@xmath53 .
the effective hoppings @xmath54 between nearest - neighbor ( @xmath55 ) mn(@xmath4 ) orbitals are smaller in sr@xmath0mnmoo@xmath1 ( in comparison with mno ) , as they are mediated by much longer mn - o - mo - o - mn paths .
however , @xmath56 is expected to be also smaller in sr@xmath0mnmoo@xmath1 , which is more itinerant material than mno .
thus , smaller @xmath54 in sr@xmath0mnmoo@xmath1 is compensated by smaller @xmath56 , and the resulting se interactions are expected to be comparable with those in mno .
the qualitative discussions are supported by direct calculations of inter - atomic magnetic interactions ( fig.[fig.jij ] ) . in the fm state ,
the @xmath57 exchange @xmath58 between next - nearest - neighbors in the mn sublattice is the strongest interaction .
the @xmath55 ( @xmath59 ) interaction @xmath60 is significantly weaker and becomes antiferromagnetic around @xmath39@xmath11@xmath61 .
however , the fm interaction @xmath58 largely prevails and the fm phase remains stable at least up to @xmath39@xmath11@xmath42 .
the @xmath55 interactions @xmath62 between mn and mo sublattices is small and does not play any role in the problem .
the breathing distortion in the afm state stabilizes the afm coupling ( both between first and second nearest neighbors in the mn - sublattice ) . @xmath63 and @xmath64 are of the order of 5 - 8 mev , and comparable with similar interactions in mno @xcite .
thus , within the interval @xmath48@xmath65@xmath39@xmath41@xmath42 _ both _ fm and type - ii afm phases are locally stable .
next , we argue that these two local minima of the total energy may exist in the very narrow energy range , so that _ two magnetic solutions are nearly degenerate_. this is qualitatively supported by results of direct total energy calculations shown in fig.[fig.etot ] .
the situation when the fm and afm solutions become nearly degenerate can occur roughly around @xmath39@xmath11@xmath61 .
we have argued that sr@xmath0mnmoo@xmath1 may have ( at least ) two total energy minima corresponding to the metallic fm and insulating type - ii afm states , which coexist in the very narrow energy range .
we expect the competition between these states to prevent the formation of the long - range magnetic order which features the experimental behavior of sr@xmath0mnmoo@xmath1 down to the very low temperature .
the quantitative description of such a situation , using results of first - principles electronic structure calculations , could be one of the possible extensions of the present work . | it is argued that the magnetic behavior of sr@xmath0mnmoo@xmath1 is determined by the existence of two total energy minima corresponding to the metallic ferromagnetic and insulating antiferromagnetic states , which may be nearly degenerate depending on the magnitude of the breathing distortion .
+ _ pacs : _ 71.20.be ; 71.70.gm ; 72.25.ba ; 75.30.et _ keywords : _ ordered double perovskites , double exchange and superexchange interactions , electronic structure calculations , lattice distortion * on the competition between ferromagnetic and antiferromagnetic states in sr@xmath0mnmoo@xmath1 * + i. v. solovyev + _ tokura spin superstructure project ( sss ) , + erato japan science and technology corporation , + c / o national institute of advanced industrial science and technology , + aist tsukuba central 4 , 1 - 1 - 1 higashi , tsukuba , ibaraki 305 - 8562 , japan _ |
dynamics of the quantum dot hybrid qubit is obtained from the schrieffer - wolff transformed effective hamiltonian in terms of effective exchange coupling interaction between pairs of electrons .
spin dynamics in quantum dot have attracted wide attention in the scientific community both from experimental@xcite and theoretical@xcite point of view because of their long coherence times and potential scaling @xcite .
several architectures have been proposed based on single @xcite , double @xcite and triple @xcite quantum dot , later implemented in gaas @xcite , si @xcite and insb @xcite nanostructures . in view of creating an architecture
capable to assure the best compromise among fabrication , tunability , fast gate operations of one and two qubits , manipulability and scalability , hybrid qubits have been recently proposed @xcite .
they consist of a double quantum dot with three electrons distributed during the operations between the two quantum dots , with at least one electron in each .
the interest raised by this architecture is due to the possibility to obtain gate operations entirely implemented with purely electrical manipulations .
this property enables much faster gate operations than using ac magnetic fields , inhomogeneous dc magnetic fields or mechanisms based on spin - orbit coupling .
the heisenberg interaction which is the dominant mechanism of interaction between adjacent spins suffice for all the one- and two - qubits operations , removing the need of an inhomogeneous field which is required for instance in single - triplet qubits @xcite .
not so surprisingly , such an architecture grants the same symmetries in spin space as the triple dot qubit proposed in ref.@xcite , but it is simpler and more compact to fabricate , as it requires only two dots instead of three . starting from the hubbard - like model
, we derive a general effective hamiltonian for the hybrid qubit in terms of the spin operators of the three electrons and of their interactions , by defining a suitable projection operator that follows the method of schrieffer and wolff @xcite .
the method here adopted enables us to obtain analytically the coupling constants between pairs of electrons under lesser restrictive conditions than those obtained in ref.@xcite , preserving an explicit dependence of all the parameters involved .
the system is irrespective to the host material , whose properties are absorbed in the coupling coefficients .
next , dynamics generated by the interaction terms is expressed .
the heisenberg interaction should permit very fast gate operation , however by itself does not provide a universal gate .
it can not generate any arbitrary unitary transformation on a collection of spin-1/2 qubits as it allows only rotations among states with the same quantum numbers .
however , defining opportunely coded qubit states the heisenberg interaction alone is universal .
the paper is organized as follows .
section ii is devoted to the derivation of the effective hamiltonian for a single hybrid qubit and its dynamical behaviour , once realistic initial conditions are provided and a realistic implementation in silicon is studied in stationary conditions . in section
iii the pictorial representation of the eigenvectors in the bloch sphere analyzing two limiting cases of interest , corresponding to the situation in which two electrons are confined in the right or in the left dot , is presented .
this section describes the derivation of an effective hamiltonian that accounts for the quantum behaviour of the hybrid qubit in the regime of low energy excitations .
next , the dynamical behaviour in the logical subspace is analyzed .
the hybrid qubit , whose energy profile is shown schematically in fig . [ hybrid ] , is composed of a double quantum dot with two electrons in the left dot and one electron in the right dot .
levels denoted by @xmath0 and @xmath1 refer to the lowest single particle energies of the doubly occupied dot in the left , @xmath2 is the lowest single particle energy of the single occupied dot on the right .
each level has a two - fold degeneracy . the hubbard - like hamiltonian @xcite model of the double dot hybrid qubit
is written in terms of the creation and annihilation fermionic operators of the three spins , which are @xmath3 and @xmath4 respectively , and consists at first order of four contributions , given by @xmath5 with @xmath6 where @xmath7 the first two terms @xmath8 and @xmath9 describe respectively the single electron energy level of each dot and the tunneling energy between the two dots .
these two terms together constitute the single - particle part of the hamiltonian .
the last two terms @xmath10 and @xmath11 constitute instead the two - body part , i.e. , the coulomb interaction between pairs of electrons belonging to the same dot ( intra - dot interaction ) or to different dot ( inter - dot interaction ) . taking into consideration only the two - body part in correspondence to @xmath10
it was observed only an anti - ferromagnetic exchange coupling between the pair of electrons @xcite .
numerical results based on heitler - london approximation or hund - mulliken molecular - orbit method have instead shown that the exchange coupling can be ferromagnetic under certain parameters conditions @xcite . here , we choose to start our analysis from this exact second quantized hamiltonian ( [ hubbard ] ) to express exchange couplings between pair of electron spins .
in particular the coefficients here introduced in eq.([j ] ) are : the spin - exchange @xmath12 , the pair - hopping @xmath13 and the occupation - modulated hopping terms @xmath14 .
these parameters , as well as the coulomb energies and the intra - dot bias voltage , are fixed by the geometry of the system . on the contrary , the tunneling parameters and the inter - dot bias voltage are tunable , by acting on the potential barriers .
the hamiltonian ( [ hubbard ] ) is diagonalized in the restricted hilbert space defined by slater determinants constructed from orthogonalized one - electron wave functions , that give an accurate low - energy spectrum .
it is well known that higher - lying base states may be accounted for by perturbation theory , for example in lattice model .
we shall not attempt such a reduction here for the quantum dot problem , but we merely point out its main effect would be a renormalization of the parameters involved .
the high - energy states of the hamiltonian are eliminated to yield an effective spin hamiltonian using degenerate perturbation theory or a canonical transformation , such as the schrieffer - wolff transformation . in the same spirit of ref .
@xcite , by emplyoing the projector operator method @xcite , we convert the hubbard - like hamiltonian ( [ hubbard ] ) into its effective equivalent with spin - spin interactions , i.e. exchange coupling , between pairs of electrons .
we define the projector @xmath15 and its complementary @xmath16 , @xmath17\qquad q=1-p,\ ] ] in such a way that the total hilbert space of the three electrons system is divided into a subspace @xmath15 , which consists of holes and single occupancies and the complementary subspace @xmath16 with at least one double occupancy .
we point out that a system with only spin degeneracy is considered .
this holds for both iii v semiconductors for which direct bandgap sussists , as well as valley degenerate indirect bandgap semiconductors for which the degeneracy is lifted in nanoscale devices @xcite .
it is possible to demonstrate starting from the schrdinger equation @xcite that @xmath18 describes as well the dynamics of @xmath19 .
in addition we observe that the operator @xmath16 can be expanded as @xmath20 and concerning with the low energy excitation the term @xmath21 could be replaced by @xmath22 , where @xmath23 is the energy difference of the energy with one @xmath24 and the energy of @xmath25 .
finally the effective hamiltonian appears in the following form @xmath26 to calculate all the contributions in eq.([eff ] ) , let s consider initially all the possible configurations with single and double occupancies in the three orbitals .
the configurations are shown in fig .
[ inter ] and are divided in different boxes with different colours in such a way to have at our disposal a visualization of the terms that contribute in a consistent way to the calculations , neglecting those with higher energy . .
the configuration in the blue box at the left is the one of reference and correspond to the projection p. the configurations in the blue box on the right , having the energy of the same order of magnitude of the configuration on the left , contribute in a consistent way to the effective hamiltonian .
those in the red box in the upper part have higher energies and are neglected.,scaledwidth=45.0% ] for completeness we provide the explicit expression of the energies associated to each configuration : @xmath27 where the indices @xmath28 , assuming only integer values , denote the number of electrons in each energy level .
the quantities of interest , appearing in eq.([eff ] ) are the differences of @xmath29 with the energy associated to @xmath25 , corresponding to the case in which there is only one electron in each orbital , that is @xmath30 in order to separate the main contribution from perturbations at this point we evaluate the order of magnitude of the energies involved : the inter - dot energies indeed are smaller with respect to the intra - dot ones , that is @xmath31 @xmath32 and @xmath33 are tunable .
this condition allows us to neglect some terms with respect to others since they contribute less .
in particular the neglected terms are those in correspondence to the highest energy differences ( depicted in the red box in fig .
[ inter ] ) that are of the order of magnitude of @xmath34 .
under such conditions , the contributions appearing in eq.([eff ] ) return : @xmath35 @xmath36 where the following identities , that directly link the fermionic operators with the spin operators , have been used : @xmath37-\bold{s}_k\cdot\boldsymbol{\sigma}_{\sigma\sigma'},\end{aligned}\ ] ] where the spin operator is defined by @xmath38 and @xmath39 is the pauli operator .
collecting finally all the terms and upon neglecting the constant factors , we finally reach the closed form for the effective hamiltonian @xmath40 where , neglecting the configurations @xmath41 and @xmath42 , that give the highest energy differences , the effective coupling constants are given by @xmath43 the exchange couplings include the effects of dot tunneling , dot bias , and both on - site and off - site coulomb interactions . each coupling term contains the term @xmath44 which is the ferromagnetic direct exchange between the two electrons from their coulomb interactions and the anti - ferromagnetic superexchange .
consequently , the value of each @xmath45 can be either positive or negative and it depends stricly on the values of the parameters entering into the expressions ( [ coupling ] ) .
the coupling constants are tunable thanks to the control on the tunneling couplings @xmath46 which can be provided by external gates and on the inter - dot bias voltage @xmath47 . on the contrary , the coulomb energy @xmath48 , @xmath49 , @xmath50 and @xmath51 , as well as the intra - dot bias voltage @xmath52 , are geometry dependent and they can not be easily tuned .
these approximations reflect the realistic conditions in which a quantum dot is operated .
we point out that if condition in eq.([eps ] ) is neglected , the effective couplings ( [ coupling ] ) maintain the same structure and the same dependence on the physical parameters , such as the tunneling , but contain also the energy associated to the intermediate states in the red box in fig .
[ inter ] . in this analysis
we have neglected the overhauser field from the lattice nuclear spins .
it is neglegible for fast operations or for purified silicon , for which the hyperfine coupling can be neglected with respect to the strong exchange coupling between pairs of spins . in all the other cases a more detailed analysis including the nuclear field operator in each dot
is needed .
we have reached a closed form for the effective hamiltonian describing the hybrid qubit constituted by only direct exchange interactions between pair of spins .
the dependence of all the physical parameters of the dots is preserved . in this paragraph
the effective hamiltonian is recast in a more compact and useful form using the logical basis introduced in ref.@xcite . with the aim of defining the logical basis for the hybrid qubit , we first enumerate the possible transitions between the three electrons spin states that can be induced by manipulations which preserve total spin angular momentum .
to do this we consider the hilbert space of three electron spins that has a dimension of @xmath53 .
the total spin eigenstates form a quadruplet with @xmath54 and @xmath55 and two doublets each with @xmath56 and @xmath57 , where the square of the total spin is @xmath58 and the @xmath59-component of the total spin is @xmath60 . to encode our qubit we restrict to the two - dimensional subspace of three - spin states with spin quantum numbers @xmath56 and @xmath61 , like in ref.@xcite .
we point out that only states with the same s and @xmath62 can be coupled by spin independent terms in the hamiltonian .
the logical basis @xmath63 , that we are going to introduce , is constituted by singlet and triplet states of a pair of spins , for example the pair in the left dot , in combination with the angular momentum of the third spin , localized in the right dot .
this means that the logical states choosen are finally expressed in this way @xmath64 , @xmath65 and @xmath66 are respectively the singlet and triplet states , whose explicit form , in terms of the eigenstates of @xmath67 , is here reported for completeness @xmath68 explicit calculations of the matrix elements of @xmath69 in this basis give @xmath70 leading directly to the following matrix form @xmath71 the effective hamiltonian just derived ( [ effmatrix ] ) could be successfully exploited to analyze the dynamical behaviour of the system .
eq.([effmatrix ] ) contains indeed all the informations on the system through the coupling constants , that when varied from the external allow operations on the qubit .
gate operations are implemented by acting on the energy splitting and by increasing the tunnel couplings , which drive transitions between qubit states .
we point out that matrix elements of the hamiltonian depend on the tunnel amplitude and are inversely proportional to the energy difference .
we conclude this section by presenting an example of the dynamical evolution of the system by starting from a specific initial condition .
we start from the initial condition in which the qubit is prepared in the state denoted by @xmath72 in the logical basis . in appendix
[ a ] the full exact dynamics is derived and the explicit probability amplitudes concerning the state of the system at the time instant @xmath73 are given , once the initial condition is fixed .
exploiting eqs.([10 ] ) we obtain the condition on the time instant under which the qubit undergoes a transition from the logical state @xmath72 to @xmath74 @xmath75 where the coefficients @xmath76 , @xmath77 and @xmath78 are defined in eqs.([abc ] ) .
we notice that the @xmath79 function is defined in the interval @xmath80 $ ] , as a consequence there is only one condition for the coefficients of eq.([t ] ) returning real valued time , that is @xmath81 the condition for the coefficients in terms of physical parameters of the system is expressed by @xmath82 this relationship gives a simple condition that links directly the coupling constants involved in the system .
it is consequently possible to change suitably the tunnel barrier between the dots to guide the system toward a switch from @xmath72 to @xmath74 . in this subsection
we study an example of the implementation of the hybrid qubit in silicon . differently from bulk silicon which has a conduction band structure with
three couples of energy degenerate valleys , in silicon quantum dots a sequence of energy levels is present with their valley degeneracy removed by interface delta - like potentials@xcite . in the past ,
some of the authors already explored the creation of robust qubits in silicon and in particular it has been applied the current spin density functional theory ( cs - dft ) to investigate the geometrical effects on the filling of the lowest valley states in quantum dot formed in single electron transistors ( set ) @xcite .
the set was modeled as a silicon nanowire on the top of a silicon dioxide slab where a tri - sided metal gate , insulated from the nanowire with a silicon dioxide layer and from the source / drain contacts with two silicon nitride spacers , electrostatically forms a quantum dot , like in ref.@xcite . by using this model ,
the effects of the fabrication variability of the sizes of the device on the valley filling and addition energy patterns were studied .
it has been shown in ref.@xcite that variations of width and thickness of the nanowire with squared sections entail important changes in the valley filling sequences , whereas the variability of the gate length does not imply a significant modification of the filling patterns .
the filling of states belonging only to one type of valley is possible for the first electrons if the silicon thickness is much smaller than the width of the nanowire . as a result
, a high valley orbit energy can be obtained so that higher quantum dot energy levels related to other valleys can be neglected leaving a single fundamental valley .
the silicon 2quantum dot system , studied in this section , is well described by the single valley effective hamiltonian ( 19 ) if we assume a high valley orbit energy in both quantum dots .
focusing on the stationary solutions of the hamiltonian , in fig .
[ fig : eigen - t2x2 ] we report the eigenvalues and the probabilities of the eigenvectors to be found in @xmath72 as a function of the tunneling couplings @xmath83 and @xmath84 . for the simulations most of the parameters of the effective hamiltonian
are taken from ref.@xcite , whereas @xmath51 and @xmath50 have been increased because they were slightly underestimated as stated in ref.@xcite .
the system is highly sensitive to the choice of those parameters and we set them to have a good control over the eigenstates of the system in the range of the tunnelling couplings considered .
the energy gap between the eigenvalues is small for low tunneling couplings whereas it increases for higher @xmath83 and @xmath84 values .
looking at the first eigenvector , the probability to be in @xmath72 is closer to 1 when @xmath85 and @xmath83 are low .
increasing @xmath85 and @xmath83 the probability to be in @xmath72 instead decreases to 0 in such a way that the first eigenvector is closer to @xmath74 state .
the opposite behaviour is observed for the second eigenvector . on the contrary ,
if exactly the same parameters values of ref.@xcite are used , the tunnelling couplings are not so effective in controlling the eigenstates of the system , leaving the lowest eigenvector close to @xmath72 ( not shown ) . as a function of the tunneling couplings @xmath83 and @xmath84 for the first eigenstate ( left ) and the second one ( right ) .
the parameters used in the simulation are : @xmath86=0 mev , @xmath87=0.3 mev , @xmath88=0.35 mev , @xmath34=9.8 mev , @xmath89=9.8 mev , @xmath90=11 mev , @xmath91=1.8 mev , @xmath92=1.8 mev , @xmath93=9.8 mev , @xmath94=0.3 mev , @xmath95=0.3 mev , @xmath96=0.3 mev , @xmath97=0.5 mev , @xmath98=0.7 mev and @xmath99=0.7 mev.,scaledwidth=50.0% ] the read out of the qubit state coincides with the read out of the spin state of electrons confined in the doubly occupied quantum dot .
a set can be used to electrostatically sense the spin state .
the set and the quantum dot are coupled through an energy barrier controlled electrostatically by a gate . during manipulation time interval ,
the electrostatic barrier between the doubly occupied quantum dot and the set is high thus both the wavefunctions of the electrons do not interact , causing a negligible effect on the spin state of the electron couple .
when read out starts , tunneling is allowed from the doubly occupied quantum dot to the set by a reduction of the interposed electrostatic barrier .
when the couple of electrons are in a singlet state the corresponding wavefunction is more confined and the tunneling rate to the set is lower than that of the triplet state which has a broader wavefunction .
when the electron tunnels , the electrostatic potential landscape changes and so does the current passing through the electrostatically coupled set @xcite .
the measurement of the time interval between the read out signal and the current variation in the set reveals the spin state of the electron couple .
a high number of electrons have to be collected in the set to mix its collective spin state , reducing the perturbing effect of set / quantum dot mutual exchange interactions on the spin state of the doubly occupied quantum dot . in this way ,
the final read out result is mostly determined by the spin state of the electron couple in the quantum dot .
the objective of this section , exploiting the effective hamiltonian ( [ heff ] ) obtained in section ii , is to analyze how exchange interaction can drive arbitrary qubit operations .
eigenstates are here derived considering the full basis of the three electrons and represented as points on the bloch sphere in connection with the states of the logical basis presented earlier . the single qubit operation is directly linked to the variation of the effective coupling , that appears as a rotation around a specific axis on the bloch sphere .
eigenstates and eigenenergies of the three electron spins in two dots are shown in eqs.([eig ] ) .
as we have already mentioned , the eight spin eigenstates of the hamiltonian comprise a quadruplet @xmath100 and high- and low - energy doublets @xmath101 and @xmath102 , @xmath103 the relative energies are given by @xmath104 where @xmath105 . at this point
we focus our attention on some cases of physical interest .
the cases under study correspond to two different situations : one in which the coupling @xmath106 is greater with respect to the others , which means that two electrons are confined in the right dot , and the opposite situation in which two electrons are in the left dot , that means a larger contribution from @xmath107 .
the explicit expressions of eigenvectors and eigenvalues in such limiting cases are collected in appendix [ b ] .
states of the qubit correspond to points on the bloch sphere shown in fig .
[ bloch ] .
exchange @xmath106 drives rotations about the vertical axis , where @xmath108 and @xmath109 , given in eqs.([b1 ] ) , denote respectively states with the pair of spins in the right dot forming a singlet and a triplet states . on the other hand exchange @xmath107 drives rotations about an axis tilted by 120 and defined by doublets @xmath110 and @xmath111 ,
see eqs.([b3 ] ) , that correspond to singlets and triplets states on the left dot .
arbitrary single - qubit operations can be achieved by concatenating up to four exchange pulses as shown in ref.@xcite .
the logical states introduced in eq.([01 ] ) are equivalent to @xmath112 in the bloch sphere .
the logical basis is formed from two states with equal @xmath62 , one taken from each doublet .
the implementation of full single qubit operation needs the preparation and read out of only two of the three electron spins .
we point out that the system under study and the linear triple dot architecture proposed in ref.@xcite , due to the fact that contain three interacting electrons , have the same symmetries , providing an analogue picture in the bloch sphere .
this means that it is possible to define gate operations like in the triple dot model .
however due to the different architectures , the hamiltonian models present some distinctions .
first of all in the triple dot it is not allowed to spins in the external dots to interact directly , as their interaction is indeed mediated by the spin in the middle . in addition the parameters involved in the double dot are influenced by the confinement of two electrons in the same dot , as occurs for @xmath113 .
these differences determine a change in the expressions for eigenvalues and eigenvactors , and leave unchanged the shape .
the convenience in adopting the double dot model with respect to the triple dot rely on a great advantage from a practical point of view , requiring the control of two dots instead of three .
starting from the generalized hubbard model , we analytically derive in the schrieffer - wolff transformation framework a complete formula for the exchange coupling between three confined electrons in a double quantum dot , independently of the host material .
the tunable exchange couplings depend explicitly on intradot and interdot coulomb repulsions , tunnel couplings and interdot bias .
the three interacting electrons are used to encode a logical qubit , that combine singlet - triplet states of the pair of spins in the left dot with single spin states of the electron spin in the right dot .
we have shown how the exact dynamics of the system is exploited to guide the system towards specific states of interest , by suitably tuning the external bias voltages and the tunneling parameters .
moreover the effective hamiltonian is exploited to describe a realistic silicon qubit studying the stationary solutions .
we have also provided a picture of the system on the bloch sphere to link the tuning of the effective exchange coupling with the rotation of the involved state .
our results pave the way towards two - qubit operations controlled by the effective exchange couplings , in order to implement complex logic gates .
this work is partially supported by the project qudec , granted by the ministero della difesa of italy .
time - dependent schrdinger equation for the hybrid qubit described by hamiltonian ( [ effmatrix ] ) is here solved .
the state of the system at the initial time @xmath114 is written as a normalized superposition of the states of the logical basis @xmath63 with probability amplitudes given by @xmath115 and @xmath116 .
the normalization condition @xmath117 is satisfied . due to the conservation of the total angular momentum operator
, it follows that also at a generic time instant @xmath73 , the state of the system can be written analogously with probability amplitudes @xmath118 and @xmath119 depending explicitly on time @xmath120 by inserting this expression into the time - dependent schrdinger equation @xmath121 and by solving the system of two first order differential equations for the probability amplitudes @xmath118 and @xmath119 , we finally obtain @xmath122 where @xmath123 and @xmath124 eqs.([ab ] ) contains the more general form for the probability amplitudes at every time instant @xmath73 .
once that the initial condition is fixed it is possible to extract the values for the coefficients @xmath125 and @xmath126 . in the case of the specific initial condition analyzed in section ii in which the system is prepared in the state of the logical basis corresponding to @xmath127 ,
the coefficients are @xmath128 after straightforward calculations we get the probability amplitudes @xmath129\\ b(t)=-ie^{i\alpha t}\frac{(a-\alpha)^2-\beta^2}{\beta c}\sin(\beta t).\\ \end{array } \right.\ ] ]
in this appendix eigenvectors and eigenvalues of the hybrid qubit , decribed by the effective hamiltonian ( [ heff ] ) , are presented in two special cases .
two limiting conditions of interest from the practical point of view , are analyzed .
1 . case @xmath130 + under the condition on the exchange coupling @xmath130 , that means that two electron are confined in the right dot , eigenvectors and eigenvalues in eqs.([eig ] ) and ( [ eigl ] ) become @xmath131 @xmath132 2 .
case @xmath133 + on the other hand , the opposite condition corresponding to two electrons confined in the left dot , that is @xmath133 , gives as eigenvectors and eigenvalues @xmath134 @xmath135 26 m.d .
shulman , o.e .
dial , s.p .
harvey , h. bluhm , v. umansky , and a. yacoby , science * 336 * , 202 ( 2012 ) .
johnson , j.r .
petta , j.m .
taylor , a.yacoby , m.d .
lukin , c.m .
marcus , m.p .
hanson , and a.c .
gossard , nature ( london ) * 435 * , 925 ( 2005 ) .
koppens , j.a .
folk , j.m .
elzerman , r. hanson , l.h .
willems van beveren , i.t .
vink , h.p .
tranitz , w. wegscheider , l.p .
kouwenhoven , l.m.k .
vandersypen , science * 309 * , 1346 ( 2005 ) .
maune , m.g .
borselli , b. huang , t.d .
ladd , p.w .
deelman , k.s .
holabird , a.a .
kiselev , i. alvarado - rodriguez , r.s .
ross , a.e .
schimitz , m. sokolich , c.a .
watson , m.f .
gyure , and a.t .
hunter , nature ( london ) * 481 * , 344 ( 2012 ) .
h. bluhm , s. foletti , i. neder , m. rudner , d. mahalu , v. umansky , and a. yacoby , nat .
* 7 * , 109 ( 2011 ) .
a.m. tyryshkin , s. tojo , j.j.l .
morton , h. riemann , n.v .
abrosimov , p. becker , h .- j .
pohl , t. schenkel , m.l.w .
thewalt , k.m .
itoh , and s.a .
lyon , nat .
mater . * 11 * , 143 ( 2012 ) .
rui li , xuedong hu and j.q .
you , phys .
b * 86 * , 205306 ( 2012 ) .
w.a . coish and daniel loss , phys .
b * 72 * , 125337 ( 2005 )
. shun - qing shen and z.d .
wang , phys .
b * 61 * , 9532 ( 2000 ) .
d. loss and d.p .
divincenzo , phys .
a * 57 * , 120 ( 1998 ) .
divincenzo , d. bacon , j. kempe , g. burkard , and k. whaley , nature ( london ) * 408 * , 339 ( 2000 ) .
taylor , h .- a .
engel , w. dr , a. yacoby , c.m .
marcus , p. zoller , and m.d .
lukin , nat . phys . * 1 * , 177 ( 2005 ) . j. levy , phys . rev . lett . *
89 * , 147902 ( 2002 ) .
petta , a.c .
johnson , j.m .
taylor , e.a .
laird , a. yacoby , m.d .
lukin , c.m .
marcus , m.p .
hanson , and a.c .
gossard , science * 309 * , 2180 ( 2005 ) .
kikkawa and d.d .
awschalom , phys .
* 80 * , 4313 ( 1998 ) .
s. amasha , k. maclean , iuliana p. radu , d.m .
zumbhl , m.a .
kastner , m.p .
hanson , and a.c .
gossard , phys .
. lett . * 100 * , 046803 ( 2008 ) .
koppens , k.c .
nowack , and l.m.k .
vandersypen , phys .
. lett . * 100 * , 236802 ( 2008 ) . c. barthel , j. medford , c.m .
marcus , m.p .
hanson , and a.c .
gossard , phys .
* 105 * , 266808 ( 2010 ) .
a.m. tyryshkin , s.a .
lyon , a.v .
astashkin , and a.m. raitsimring , phys . rev .
b * 68 * , 193207 ( 2003 ) .
a. morello , j.j .
pla , f.a .
zwanenburg , k.w .
chan , k.y .
tan , h. huebl , m. mttnen , c.d .
nugroho , c. yang , j.a .
van donkelaar , a.d.c .
alves , d.n .
jamienson , c.c .
escott , l.c.l .
hollenberg , r.g .
clark , and a.s .
dzurak , nature ( london ) * 467 * , 687 ( 2010 ) . c.b .
simmons , j.r .
prance , b.j .
van bael , t.s .
koh , z. shi , d.e .
savage , m.g .
lagally , r. joynt , m. friesen , s.n .
coppersmith , and m.a .
eriksson , phys .
lett . * 106 * , 156804 ( 2011 ) .
xiao , m.g .
house , and h.w .
jiang , phys .
* 104 * , 096801 ( 2010 ) .
j. w. g. van den berg , s. nadj - perge , v. s. pribiag , s. r. plissard , e. p. a. m. bakkers , s. m. frolov , and l. p. kouwenhoven , phys .
lett . * 110 * , 066806 ( 2013 ) .
z. shi , c.b .
simmons , j.r .
prance , j.k .
gamble , t.s .
koh , y .- p .
shim , x. hu , d.e .
savage , m.g .
lagally , m.a eriksson , m. friesen and s.n .
coppersmith , phys .
* 108 * , 140503 ( 2012 ) .
schrieffer and p.a .
wolff , phys . rev . * 149 * , 491 ( 1966 ) . j.h .
jefferson , and w. husler , phys .
b * 54 * , 4936 ( 1996 ) .
g. burkard , d. loss , and d.p .
divincenzo , phys .
b * 59 * , 2070 ( 1999 ) .
xuedong hu and s.das sarma , phys .
a * 61 * , 062301 ( 2000 ) .
m. de michielis , e. prati , m. fanciulli , g. fiori , and g. iannaccone , applied physics express * 5 * , 124001 ( 2012 ) .
m. friesen and s.n .
coppersmith , phys .
rev b * 81 * , 115324 ( 2010 ) .
xin wang , shuo yang , and s.das sarma , phys .
b * 84 * , 115301 ( 2011 ) .
e. prati , m. de michielis , m. belli , s. cocco , m. fanciulli , d. kotekar - patil , m. ruoff , d. p kern , d. a. wharam , j. verduijn , g. c. tettamanzi , s. rogge , b. roche , r. wacquez , x. jehl , m. vinet and m. sanquer , nanotechnology * 23 * , 215204 ( 2012 ) . | quantum dot hybrid qubits formed from three electrons in double quantum dots represent a promising compromise between high speed and simple fabrication for solid state implementations of single qubit and two qubits quantum logic ports .
we derive the schrieffer - wolff effective hamiltonian that describes in a simple and intuitive way the qubit by combining a hubbard - like model with a projector operator method . as a result ,
the hubbard - like hamiltonian is transformed in an equivalent expression in terms of the exchange coupling interactions between pairs of electrons .
the effective hamiltonian is exploited to derive the dynamical behaviour of the system and its eigenstates on the bloch sphere to generate qubits operation for quantum logic ports . a realistic implementation in silicon and the coupling of the qubit with a detector
are discussed . |
it is well known that the symmetry - adapted solution of the nonlinear hartree - fock ( hf ) equations of an electronic system is sometimes unstable .
an unstable solution corresponds to a saddle point of the energy as a function of the orbital parameters , and breaking of space and/or spin symmetries of the wave function then necessarily leads to one or several lower - energy hf solutions .
the stability conditions of the hf equations were first formulated by thouless @xcite , and the different instabilities were first categorized by iek and paldus @xcite . for closed - shell systems
, one may encounter `` singlet instabilities '' when only spatial symmetry is broken , and `` triplet ( or nonsinglet ) instabilities '' when spin symmetry is also broken .
there is thus a symmetry dilemma @xcite in choosing between the symmetry - adapted wave function of higher hf energy and a symmetry - broken wave function of lower hf energy , in particular as a reference for a post - hartree - fock calculation . a clear example is provided by closed - shell hydrogen rings h@xmath0 with equal bond lengths @xcite ( see , also , ref . ) .
the symmetry - adapted hf solution exhibits singlet instabilities for sufficiently large numbers of hydrogen atoms , and one can obtain symmetry - broken hf solutions with orbitals localizing on either the atoms or the bonds .
however , both mller - plesset perturbation theory and linearized coupled cluster doubles theory ( also called cepa0 or dmbpt@xmath1 ) give a lower total energy when starting from the symmetry - adapted solution than when starting from the symmetry - broken solutions , which casts doubts on the physical significance of the symmetry - broken solutions .
of course , the symmetry dilemma would be removed with a full configuration - interaction calculation which must give one unique solution , independent of the orbitals used .
quantum monte carlo ( qmc ) approaches are alternatives to the traditional quantum chemistry methods @xcite .
the two most commonly used variants are variational monte carlo ( vmc ) which simply evaluates the energy of a flexible trial wave function by stochastic sampling , and fixed - node diffusion monte carlo ( dmc ) which improves upon vmc by projecting the trial wave function onto the ground state subject to the condition that the nodes of the projected wave function are the same as those of the trial wave function . for large systems ,
the most common form of the trial wave function is a jastrow factor multiplied by a fixed hf determinant ( though for small systems one can do much better by replacing the hf determinant by a linear combination of optimized slater determinants ) .
if a system exhibits hf instabilities , then qmc also faces the symmetry dilemma in choosing between different hf wave functions .
indeed , different hf wave functions necessarily lead to different energies not only in vmc , but also in dmc since the nodes of these hf wave functions are generally different .
this symmetry dilemma in dmc is only due to the fixed - node approximation , since without this approximation dmc would give one unique solution , independent of the trial wave function .
of course , for systems that are not very large , symmetry breaking could probably be avoided in the first place by optimizing the orbitals within vmc @xcite , instead of using fixed hf orbitals . in this work , we study the impact of the hf symmetry dilemma for qmc in hydrogen rings h@xmath0 . in sec .
ii , we review the hf symmetry - breaking problem in these systems , and discuss the effect of using a slater basis versus a gaussian basis . in sec .
iii , we explain the qmc methodology and report our vmc and dmc results .
our conclusions are summarized in sec .
in previous studies @xcite , the electronic structure of periodic rings of @xmath2 evenly spaced hydrogen atoms ( with a fixed distance of @xmath3 ) has been investigated .
the number of hydrogen atoms is restricted to @xmath2 in order to obtain a possible closed - shell single - determinant solution with @xmath4 occupied orbitals .
the symmetry - adapted hf wave function has a metallic character ( in the limit of an infinite ring ) and can be expressed with either delocalized canonical orbitals or localized wannier orbitals ( which do not have an exponential decay ) .
the canonical orbitals , except for the lowest one , are doubly degenerate , and in a minimal basis the orbital coefficients are fixed by the cyclic symmetry . besides the symmetry - adapted ( sa ) solution , two different symmetry - broken hf solutions of lower energy can be obtained beyond critical ring sizes , when using unit cells of 2 hydrogen atoms .
one solution corresponds to orbitals localized on hydrogen atoms and is referred to as the symmetry - broken atom - centered ( sb - ac ) solution , while the other corresponds to orbitals localized on bonds and is referred to as the symmetry - broken bond - centered ( sb - bc ) solution .
the sb - bc solution has the lowest hf energy .
the three solutions can be schematically described as @xmath5h@xmath5h@xmath5 , @xmath5h@xmath6h@xmath7 , and @xmath5h h@xmath5 . in each case , the symmetry breaking is accompanied by an opening of an energy gap between occupied and virtual orbitals , and orbitals decay much more rapidly than for the symmetry - adapted solution @xcite , in agreement with the theoretical result of kohn @xcite . in order to distinguish the three different wave functions ,
one may look at the one - particle density matrix @xmath8 @xmath9 containing the coefficients @xmath10 of the occupied molecular orbitals @xmath11 , @xmath12 expanded in a minimal set of atom - centered basis functions , i.e. one single basis function @xmath13 per hydrogen atom . as depicted in figure [ densmat ] , for the sa solution , we see equal elements on the diagonal and the sub - diagonals of the density matrix . for the sb - ac solution ,
an alternation of element values on the diagonal of the density matrix is obtained , but equal elements on the first sub - diagonal , and for the sb - bc solution we have equality of the diagonal elements and alternation on the first sub - diagonal . in ref . ,
a minimal gaussian basis set ( five @xmath14 gaussian functions contracted to one single basis function for each hydrogen atom ) was used . however , gaussian basis functions are not appropriate for all - electron qmc calculations .
they give large statistical fluctuations due to their incorrect vanishing gradient at the nuclear positions .
it is thus much preferable to use slater basis functions which correctly have a non - zero gradient on the nuclei and an exponential decay at large distance . in this work
, we use a minimal slater basis set ( one @xmath15 slater function on each hydrogen atom ) with an exponent of 1.17 , which is smaller than the optimal exponent of 1.24 for an isolated h@xmath16 molecule .
spin - restricted hf ( and mp2 ) calculations were performed with an experimental code for ring systems , employed already for the previous studies @xcite .
the necessary integrals over slater functions have been calculated with the program smiles @xcite . in order to obtain the symmetry - broken hf solutions , we start from a set of localized wannier orbitals describing either an ionic situation or an explicit bond in the two - atom unit cell , and use an iterative configuration interaction procedure using singly excited determinants @xcite instead of diagonalizing a fock operator . [ cols="<,^,^,^,^ " , ] [ tab : hfmp2vmcdmc ] figure [ dmc ] shows the corresponding dmc results .
the energy ordering is the same as in vmc and mp2 , the sa wave function giving the lowest dmc total energy , and thus the smallest fixed - node error .
as shown in table [ tab : hfmp2vmcdmc ] , the energy splittings between the different solutions are much smaller in dmc .
this indicates that dmc is less sensitive to symmetry breaking than other correlation methods .
it is an interesting feature for cases where symmetry breaking can not be avoided . of course ,
symmetry breaking would probably be avoided if the orbitals were reoptimized in the presence of the jastrow factor within vmc , but this is computationally expensive for large systems .
when hf trial wave functions are used in qmc calculations , in case of hf instabilities qmc faces the hf symmetry dilemma in choosing between the symmetry - adapted solution of higher hf energy and symmetry - broken solutions of lower hf energies . in this work ,
we have examined the hf symmetry dilemma in hydrogen rings h@xmath0 which present hf singlet instabilities for sufficiently large ring sizes .
we have shown that using a slater basis set , instead of a gaussian basis set , delays the onset of hf symmetry breaking until larger rings and slightly reduces the energy splittings between the symmetry - adapted and symmetry - broken wave functions .
when using these different hf wave functions in vmc and dmc , we have found that the energy ordering is reversed ; the symmetry - adapted wave function always giving the lowest energy .
this confirms previous post - hartree - fock studies in showing that these symmetry - broken solutions are bad starting wave functions for correlated calculations .
the fact that the symmetry - adapted wave function gives the lowest dmc energy indicates that this wave function has more accurate nodes than the symmetry - broken wave functions .
the present experience thus suggests that spatial symmetry is an important criterion for selecting good trial wave functions . for systems that are not very large , the symmetry - breaking problem
could probably be avoided altogether by optimizing the orbitals within the quantum monte carlo calculation , rather than using fixed hf orbitals .
this work has been financed mainly through the deisa network , project stop - qalm .
all qmc calculations have been performed on the ibm bluegene machines in jlich and munich ( germany ) .
c.j.u . was supported in part by the nsf ( grant nos .
dmr-0908653 and che-1004603 ) .
the authors thank the staff of idris ( orsay , france ) for technical assistance to install , test and run the qmc program champ on these machines .
we also acknowledge using the slater integral code smiles ( madrid , spain ) for obtaining the hf starting wave functions .
discussions with p. gori - giorgi ( amsterdam , netherlands ) and j .- p .
malrieu ( toulouse , france ) were very helpful for the project . | when using hartree - fock ( hf ) trial wave functions in quantum monte carlo calculations , one faces , in case of hf instabilities , the hf symmetry dilemma in choosing between the symmetry - adapted solution of higher hf energy and symmetry - broken solutions of lower hf energies . in this work , we have examined the hf symmetry dilemma in hydrogen rings which present singlet instabilities for sufficiently large rings .
we have found that the symmetry - adapted hf wave function gives a lower energy both in variational monte carlo and in fixed - node diffusion monte carlo .
this indicates that the symmetry - adapted wave function has more accurate nodes than the symmetry - broken wave functions , and thus suggests that spatial symmetry is an important criterion for selecting good trial wave functions . |
current and forthcoming photometric surveys aim to image a significant fraction of the sky . in doing so
, they will obtain the angular positions of millions of galaxies . realizing the scientific potential of these surveys requires an estimate of the _ redshift distribution of the galaxies _ , which is important for connecting measurements such as tomographic weak lensing @xcite , the integrated sachs - wolfe effect and angular power spectra to the underlying cosmological model . in this work
we investigate a method to measure galaxy redshift distributions using angular cross - correlations with an overlapping spectroscopic sample .
we outline the approach as follows .
consider two galaxy datasets : a spectroscopic sample with a known redshift distribution , and a photometric sample with an unknown redshift distribution .
the samples overlap on the sky and in redshift .
since they are sampled from the same underlying density field , we expect that they will share a positive cross - correlation function regardless of galaxy attributes such as colour and luminosity
. the amplitude of the angular cross - correlation will increase with the degree of overlap of the two samples .
therefore , the redshift distribution of the photometric sample can be mapped out by dividing the spectroscopic sample into adjacent redshift bins , and for each bin measuring the angular cross - correlation with the photometric sample . in this work
we use this technique to constrain the redshift distribution of galaxies in tomographic bins within the kilo - degree survey ( kids ) using the spectroscopic 2-degree field lensing survey ( 2dflens , * ? ? ?
* ) to trace the surrounding large - scale structure .
knowledge of the redshift distribution of the source galaxies is a critical component of a weak lensing analysis because it is required to calculate the expected weak lensing signal for a given cosmological model ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
uncertainty and bias in the source redshift distribution directly propagates to derived cosmological constraints as one of the most important astrophysical systematics .
the required level of systematic error control increases in severity for future surveys : for `` stage iv '' dark energy experiments @xcite such as the large synoptic sky telescope ( lsst ) and _ euclid _ , in order to avoid a degradation of dark energy constraints by more than @xmath1 , the mean and standard deviation of the photometric redshift distribution need to be measured to an accuracy @xmath2 @xcite .
many approaches have been proposed for determining source redshift distributions .
we define _ direct calibration _ methods as those that calibrate a mapping from the flux in photometric bands to a galaxy s redshift .
template - based approaches and machine learning algorithms both lie in this category , but there are various factors that make the above level of accuracy difficult to obtain , including catastrophic photometric errors , completeness requirements for spectroscopic training samples , and sample variance @xcite .
we will discuss these factors in the subsequent section .
an alternative ` indirect calibration ' approach , which we pursue in this study , is provided by cross - correlation methods . in particular , we focus on the extension and application of the _ optimal quadratic estimation _ method proposed by @xcite ( henceforth , ) , testing this method using both simulations and data .
the outline of this paper is as follows .
( [ sec : tech ] ) introduces the strengths and weaknesses of calibration via cross - correlations , and highlights the previous work in the field . in sec .
( [ sec : data ] ) we introduce the datasets we employ in this study : the kilo - degree survey , the 2-degree field lensing survey and mock catalogues built from n - body simulations . sec .
( [ sec : parm ] ) introduces the background theory , and sec .
( [ sec : oqe ] ) describes the quadratic estimator we employ to measure the redshift distribution of galaxies .
we validate our methodology using mock catalogues in sec .
( [ sec : simtests ] ) , and present the results of applying our methodology to data in sec .
( [ sec : results ] ) .
we conclude in sec .
( [ sec : final ] ) .
the key point of distinction between direct and indirect calibration approaches is that the former requires spectroscopic redshifts for a subsample of the full photometric sample , and this subsample needs to be representative of the full sample in both colour and magnitude space @xcite .
this requires that the targetted spectroscopic sample should be highly complete , i.e. , a secure redshift needs to be measured for @xmath4 of the subsample @xcite . to achieve this level of completeness
, spectroscopic redshifts are required for faint and high - redshift galaxies that are abundant in deep imaging surveys .
in contrast , for a cross - correlation analysis one is free to target any tracer of overlapping large - scale structure ( most usefully , the brightest galaxies ) , circumventing this difficulty .
achieving a high level of spectroscopic - redshift completeness for direct calibration methods presents a significant observational challenge , as the chance of obtaining a successful redshift is dependent on an object s magnitude .
therefore , spectra are typically obtained for a non - random subsample of the target catalogue .
a useful example is the deep2 survey conducted on the deimos spectrograph at keck observatory : for the highest redshift quality class , secure redshifts were only obtained for @xmath5 of the galaxies @xcite .
considering future surveys , the severity of this problem is demonstrated by the requirement for spectroscopic follow - up suggested by @xcite : obtaining @xmath4 redshift completeness at @xmath6 ( lsst depth ) would require more than 100 nights on keck .
we note that the requirements for spectroscopic follow - up can be reduced by assigning weights to galaxies during the training phase of photometric - redshift calibration @xcite .
these weights are assigned based on the colour - magnitude phase - space distribution of both the parent photo-@xmath3 sample and the follow - up spectroscopic sample .
the result is that the weighted spec-@xmath3 sample more closely matches the photo-@xmath3 sample in colour - magnitude space .
the extent to which this approach allows one to reduce the required completeness is currently a subject of study .
the above challenges are also relevant for machine learning algorithms , as they require separate training , testing and validation samples .
similar requirements exist for template - based approaches , where even though templates can be derived synthetically in principle assessing the resulting accuracy of the photo-@xmath3 estimate requires a representative spec-@xmath3 sample , which can also aid in the construction of accurate spectral templates .
moreover , deriving a bayesian prior used in the fitting process ( e.g. , bpz * ? ? ?
* ) requires a spec-@xmath3 sample ; this prior can strongly influence the final results @xcite . _
catastrophic errors _ , photo-@xmath3 estimates @xmath7 with @xmath8 , present an additional issue for direct calibration methods .
such errors occur because , with only broad - band flux information , there exist degeneracies in galaxy colours such as the confusion between the lyman and balmer breaks .
general studies of the consequences of catastrophic errors are presented by @xcite and @xcite .
these outliers can not be mitigated by re - weighting the sample .
although photo-@xmath3 calibration by cross - correlation avoids some of the issues listed above , there remain significant challenges for this approach , some of which we outline below . * * degeneracy with galaxy bias * : cross - correlations measure the combination @xmath9 , where @xmath10 is the source galaxy bias factor and @xmath11 is the source redshift probability distribution
. therefore , galaxy bias is degenerate with the redshift distribution @xcite . calibrating the redshift - dependent galaxy bias
requires extra probes ( e.g. , galaxy - galaxy lensing ) or assumptions ( e.g. , the bias varies smoothly with redshift ) .
this likely represents the dominant issue when constraining the source redshift distribution using cross - correlations . * * cosmological dependence * : the model cross - correlation function depends on both our guess of @xmath11 and the cosmological model .
this introduces a worrying circularity , as our aim is to test the cosmological model with measurements derived using @xmath11 . * * extra source of correlations * : cosmic magnification introduces additional correlations and hence can bias measurements of the redshift distribution .
in addition to changing the shape of galaxies , lensing changes their size and brightness , promoting fainter galaxies into a magnitude - limited survey and correlating foreground and background objects @xcite . * * spec-@xmath3 coverage * : the redshift range over which the source distribution can be reconstructed is limited by the redshift and areal coverage of the spectroscopic cross - correlation samples . in this section
we summarize recent work on photometric calibration with angular cross - correlations .
* * estimators : * a number of estimators have been proposed for inferring the redshift distribution of a photometric sample using an overlapping spec-@xmath3 sample @xcite .
for example @xcite develop a quadratic estimator , while @xcite , @xcite and @xcite use maximum - likelihood approaches to infer the source distribution . *
* self - calibration : * dividing a photometric sample into redshift bins allows one to cross - correlate between bins .
this correlation allows one to determine the contamination fraction for the sample , and potentially constrain other systematic errors . *
* applications : * the methodology presented by @xcite and @xcite has been applied to estimate the redshift distributions of galaxies in the sloan digital sky survey ( sdss ) @xcite , the cosmic infrared background @xcite , the canada - france - hawaii telescope legacy survey ( cfhtls ) @xcite , infrared sources from wide - field infrared survey explorer ( wise ) and the two - micron all - sky survey ( 2mass ) , and radio sources from the faint images of the radio sky at twenty cm ( first ) survey @xcite .
additionally , @xcite present the first application of this methodology to a cosmic shear analysis .
the kilo - degree survey ( kids ) is a multi - band imaging survey designed for weak gravitational lensing analyses .
the survey is being performed at the 2.6-metre vlt survey telescope where , using the @xmath12-mega - pixel wide - field camera omegacam , images are taken in four filters @xmath13 .
kids aims to image @xmath14 square degrees of the sky down to a limiting @xmath15-band magnitude of @xmath16-band images are used for galaxy shape measurements because these are the deepest observations obtained in the best seeing conditions .
the @xmath13 bands have @xmath17 limiting magnitudes @xmath18 , respectively , in a @xmath19 aperture @xcite . ] .
the first and second data releases of kids are presented by and @xcite .
based on these catalogues , gravitational lensing science analyses of the @xmath20 square degree overlap area with the gama survey @xcite were undertaken by @xcite , @xcite , @xcite and @xcite , using matched - aperture @xmath13 colours in conjunction with bpz to derive redshift probability distributions and hence the lensing efficiencies .
we performed our analyses using the third data release of kids , separately using the @xmath15-band `` kids-450 '' @xcite and @xmath21-band `` kids-800 '' imaging datasets .
@xcite carried out a careful analysis of different methods for calibrating the source redshift distribution and settled on a direct photo-@xmath3 calibration scheme , verified by estimates based on clustering in a few square degrees of overlapping deep spectroscopic fields . @xcite
have recently presented a determination of the kids redshift distribution using small - scale cross - correlations .
we map the large - scale structure in which our photometric sample is embedded using the spectroscopic 2-degree field lensing survey .
we outline the basic properties of the survey here ; full details are presented by @xcite .
the principal aim of 2dflens is to expand the area of overlap between spectroscopic galaxy surveys and gravitational lensing imaging surveys .
this facilitates two key science goals .
first , it allows a joint analysis of lensing and galaxy redshift samples including all cross - correlation statistics ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , with different applications presented by and .
second , it allows the calibration of photometric - redshift distributions using cross - correlation techniques which we present in this paper and direct calibration techniques .
2dflens was conducted on the anglo - australian telescope over 53 nights in the 14b , 15a and 15b semesters .
the two main target classes , selected from the vst - atlas survey @xcite , comprised @xmath22 luminous red galaxies ( lrgs ) across a range of redshifts @xmath23 selected by sdss - inspired cuts @xcite , and a magnitude - limited complete sample of @xmath24 objects in the range @xmath25 to assist with direct photometric calibration of the skymapper survey . in our study we analyze the lrg sample , whose normalized redshift distribution ( after merging the multiple target classes ) is illustrated in fig .
( [ nz ] ) .
2dflens observations cover an area of @xmath26 deg@xmath27 .
the overlap area with the imaging is currently limited by the progress of kids , which is still collecting data at the time of writing .
currently , the overlap area between 2dflens and the @xmath21-band ( @xmath15-band ) kids imaging is @xmath28 deg@xmath27 ( @xmath29 deg@xmath27 ) , as shown in fig .
( [ skyplot ] ) .
we map out the photo-@xmath3 redshift distribution through cross - correlations by dividing our spectroscopic sample into independent redshift bins . in choosing the width of these bins we balanced considerations of noise in the measurements with the desire for high redshift resolution , defining 18 redshift bins of width @xmath30 in the redshift range @xmath31 .
we tested the robustness of our methodology by constructing _ synthetic galaxy catalogues _ composed of overlapping photometric and spectroscopic samples , which allowed us to compare the redshift distributions reconstructed by our algorithm to the known input distributions .
we generated these mock catalogues using the scinet light cone simulations ( slics ) series of n - body simulations @xcite which have been produced using the cubep@xmath32 m code @xcite using a wmap9+bao+sn cosmological parameter set : matter density @xmath33 , baryon density @xmath34 , hubble parameter @xmath35 , spectral index @xmath36 , and normalization @xmath37 .
the box - size of the simulations is @xmath38 mpc .
the simulations follow the non - linear evolution of @xmath39 particles inside a @xmath40 grid cube . for each simulation ,
the density field is output at 18 redshift snapshots in the range @xmath41 , which are used to construct a survey cone spanning 60 deg@xmath27 .
a spherical overdensity halo finder was executed on the particle data during the simulation run , and the resulting halo catalogues were post - processed to self - consistently sample the light - cone geometry .
we sampled our mocks from these halo catalogues , as described further in sec .
( [ sec : simtests ] ) .
in this section , we describe the redshift distribution parameterization and clustering model we adopt for our analysis . throughout
, we will assume the fiducial cosmological parameters of the slics simulations , stated above .
for readability , we will begin with a heuristic description and then move on to a more rigorous treatment .
the intent of this work is to present a novel technique for measuring the redshift probability distribution of a given photometric sample of galaxies , @xmath42 .
first , we need a method to parameterise this probability distribution .
we do this by dividing the redshift range of the sample into step - wise bins and constraining the number of galaxies @xmath43 within each bin @xmath21 , assuming that their probability distribution within the bin is constant .
so if @xmath44 is a normalized top hat filter ( @xmath45 ) and @xmath46 is the total number of photometric galaxies , then the probability distribution within each bin is @xmath47 and the total distribution is @xmath48 which is normalized such that @xmath49 .
the quantity of interest for constraining @xmath42 is the cross - power spectrum between two samples of galaxies : specifically , between the full photometric sample @xmath50 and a given redshift bin of the spectroscopic sample @xmath51 .
we label this angular galaxy cross - power spectrum @xmath52 , as a function of multipole @xmath53 .
one can estimate this quantity by decomposing the projected density field for each sample into spherical harmonic coefficients @xmath54 , @xmath55 for the photometric and spectroscopic sample , respectively .
the statistical estimate is then @xmath56 . in the remainder of this section
, we will describe how we can model this quantity using our parameterization for @xmath42 , and convert this model from a power spectrum to the measured correlation function .
using the small - angle ( or ` limber ' ) approximation @xcite , the matter cross - power spectrum between two redshift bins @xmath21 and @xmath57 , with redshift distributions @xmath58 , is @xmath59 where @xmath60 is the matter power spectrum at wavenumber @xmath61 , @xmath62 is the co - moving distance to redshift @xmath3 , @xmath63 and @xmath64 . to extend eq .
( [ eq : calp ] ) to model the galaxy - galaxy power spectrum we need to model the galaxy biases for both samples which we label @xmath65 , @xmath66 and to include a shot noise component .
we will assume a linear relationship between the galaxy and matter density field , i.e. , @xmath67 , arcmin .
this angular scale corresponds to a set of physical scales where we might expect linear galaxy bias to break down and introduce a systematic error .
however , as emphasized by , the smoothness of the weighting function implies that the fourier modes being traced are on more linear scales than expected from this simplistic conversion .
the mock catalogues provide a way for us to quantify the significance of this error . for future work
, this investigation could be extended by up - weighting linear scales ( reducing the overall constraining power ) or introducing a non - linear galaxy bias component ( at the cost of further complicating the methodology ) . ] and model the galaxy bias in step - wise bins as @xmath68 @xmath69 can be either @xmath50 or @xmath70 , indicating the photometric or spectroscopic sample , respectively , and @xmath71 is the width of the @xmath72 redshift bin .
the galaxy cross - power spectrum can then be written as @xmath73 where @xmath74 is the probability distribution of the spectroscopic sample in the @xmath72 bin and @xmath75 models the shot noise component .
we now wish to expand this expression in terms of @xmath43 . applying our parameterisations for @xmath42 , @xmath74 , @xmath65 and @xmath76 from eq .
( [ eq : p ] , [ eq : calp ] , [ eqn : bias ] ) , eq .
( [ eq : calp2 ] ) reduces to @xmath77 following a similar derivation , we compute the auto - correlations between the full photometric sample ( @xmath78 ) and between the bins of the spectroscopic sample ( @xmath79 ) as : @xmath80 assuming the limber approximation , which ignores long - wavelength modes , the covariance between non - overlapping bins is zero : thus , @xmath81 when @xmath82 . for further details see . to model the shot noise components for eq .
( [ eqn : cross1],[crossautode2],[crossautode3 ] ) we assume poisson statistics , neglecting non - poisson contributions @xcite . following this assumption the shot noise components are @xmath83 $ ] and @xmath84 $ ] , where @xmath85 is the overlap fraction between the photo-@xmath3 and spec-@xmath3 sample . for observational considerations we will switch to configuration space when applying our methodology to data .
thus , we need to transform eq .
( [ eqn : cross1],[crossautode2],[crossautode3 ] ) into configuration space . to simplify the final expressions we first transform the constant number count case of eq .
( [ eq : calp ] ) : @xmath86 now the angular galaxy auto- and cross - correlation functions can be written as @xmath87 to compute the various correlation statistics we use the public software chomp introduced by @xcite . this calculation requires as input the matter power spectrum for each redshift bin , which we model using the halofit code @xcite : the halofit parameters adopted are those fit by @xcite .
chomp computes the redshift evolution in each bin from linear theory using @xmath88 , where @xmath89 is the growth function .
this approximation is valid because the redshift bins we adopt are narrow .
we note a number of systematic modelling issues which could be improved in future analysis : * _ non - linear effects_. we measure the angular correlation function to scales @xmath90 arcmin . on such scales
non - linear effects become significant and the halofit model we adopt may become inaccurate .
we also assume linear galaxy bias . *
_ bias evolution_. the cross - correlation observable is the combination @xmath91 , such that our inference of @xmath42 must depend on the redshift evolution of galaxy bias . * _ flat @xmath92 approximation_. to derive the above equations we have approximated the redshift distributions using a step - wise parametrization , such that the redshift distribution is constant within each bin .
this approximation will break down if there are steep gradients in the redshift distribution .
in this section we outline the construction and properties of the ` quadratic estimator ' we employ to measure the redshift distribution using cross - correlations .
this work extends that presented by . by _ quadratic estimator _
we are referring to a statistical estimator of a quantity , say @xmath93 , based on a quadratic combination of the available data , @xmath94 .
for example , @xmath95 .
the symbol ( @xmath96 ) indicates a statistically - estimated quantity : a value derived directly from data rather than the true value .
quadratic estimation is particularly relevant for gaussian random fields as all the information content is contained within quadratic combinations of the data ( second - order statistics ) . to construct a quadratic estimator we need to @xmath97 define the data , @xmath98 specify the quantity we wish to estimate , and @xmath99
construct a method to combine the data that gives an estimate of the desired quantity
preferably this estimator will take advantage of all of the information content within the data , thus minimising the final variance of the inferred parameter ( such an estimator is said to satisfy the cramer-rao inequality and be _ optimal _ ) . considering each of these points in turn : * * _ the data _ * : we start by considering the spherical harmonic coefficients of the projected density fields , which we write as @xmath100 and @xmath101 for the photo-@xmath3 and spec-@xmath3 samples , respectively , where @xmath102 represents the coefficients for the @xmath72 spec-@xmath3 bin .
these coefficients are computed as follows .
first , we define @xmath103 as the projected galaxy density field , where @xmath104 indicates the angular position .
the spherical harmonic coefficients are computed by projecting the density field onto a basis of spherical harmonics @xmath105 : @xmath106 we combine these coefficients into a single data vector @xmath107 . * * the estimated quantity * : the quantity we wish to determine is the number count distribution of the photometric sample in step - wise bins , labelled @xmath108 ( following the parameterization defined above ) . * * the estimator * : we begin by writing the estimator in the most general form possible : @xmath109 , where @xmath110 is a symmetric matrix and @xmath111 is a constant @xcite .
these free parameters will be fixed by imposing various conditions on the estimator . rather than making a single estimate
, we can iterate until we are satisfied with the convergence . setting @xmath112_{\rm last}$ ] as our initial guess
, the updated estimator is @xmath113_{\rm last } + { \bf x}^t { \bf e}_i { \bf x } - c_i \ , .\ ] ] requiring that the estimator is unbiased and optimal one can solve for both free parameters . being _
unbiased _ implies that an ensemble average of the estimates converges to the input or true value , @xmath114 . being _
optimal _ implies that the estimator minimizes the variance , viz .
, @xmath115 is minimised .
the final form of the estimator is @xcite @xmath113_{\rm last } + \frac{1}{2 } \sum_j[{\bf f}^{-1}]_{ij } \bigg [ { \bf x}^{t}{\bf q}_j{\bf x } - { \rm tr}({\bf q}_j { \bf a } ) \bigg ] \ , , \label{generalestimator}\ ] ] where @xmath116 @xmath117 is the covariance matrix of the data , and its derivative is @xmath118 .
we note that implicitly @xmath119 is a function of both @xmath53 and @xmath120 , i.e. , @xmath121 .
assuming many modes are included one can approximate @xmath122 as the fisher matrix @xmath123 \ , .
\label{eqn : fish1}\ ] ] when this assumption is violated the fisher matrix will be biased by sample variance .
we do not expect this assumption to have a significant effect on our results .
note , from the previous section ( eq . [ eqn : cross1],[crossautode2],[crossautode3 ] ) , @xmath119 is known .
it is the full covariance matrix between the spec-@xmath3 and photo-@xmath3 samples , including auto - correlations . in particular , @xmath124 @xmath125 moreover , from eq .
( [ eqn : cross1 ] ) , the derivative of the off - diagonal terms is @xmath126
. a more intuitive derivation of eq . ( [ generalestimator ] ) can be found by applying the newton - raphson method to the gaussian likelihood function of @xmath127 , @xmath128 .
the basic idea is to solve , iteratively , for roots of @xmath129 : the roots indicate a maximum of the likelihood function . for this alternative derivation
we refer the reader to @xcite . in this section and the subsequent one , we present important yet rather tedious mathematics ; therefore , in the aid of readability we offer a quick summary .
the primary purpose of these sections is to present the analytic form of eq .
( [ generalestimator ] ) .
we derive this expression by computing the tensor @xmath130 and the fisher matrix @xmath122 using eq .
( [ eqn : a ] ) combined with the results from sec .
( [ sec : parm ] ) .
we re - visit this derivation , which was initially presented in , because we find a number of corrections to the final form of the quadratic estimator : we show that the most general expression for @xmath131 presented in is biased , such that , @xmath132 .
however , we note that in the limit where shot - noise dominates ( labelled the ` schur limit ' by ) their expression for the estimator becomes unbiased .
we present the updated result for the harmonic - space estimator in eq .
( [ estimatorfull ] ) and its extension to configuration - space in eq .
( [ eqn : fullconfig ] ) .
we note that none of the numerical calculations in are affected , since these require the fisher matrix and not the form of the estimator .
our revisions can be understood as follows .
first , our general form of the quadratic estimator ( i.e. , eq .
( [ generalestimator ] ) ) differs from that presented in : @xmath133_{\rm last } + \frac{1}{2 } \sum_j[{\bf f}^{-1}]_{ij } \bigg [ { \bf x}^{t}{\bf q}_j{\bf x } - { \rm tr } ( { \bf a}^{-1 } { \bf a}_{,j } ) \bigg ] \ , .}_\text{eq . ( 16 ) from \citetalias{2013mnras.433.2857 m } } \label{eqn : mc}\ ] ] importantly , eq . ( [ eqn : mc ] ) is a simplified version of eq .
( [ generalestimator ] ) .
the two expressions agree because of the relation @xmath134 . in order to simplify the derivation , neglect all derivatives of @xmath135 note , we also make this approximation . and
such terms ( @xmath136 ) occur in the expression for @xmath130 .
thus , by dropping these terms , one is making an approximation of @xmath130 , @xmath137 .
this approximation breaks the relation that equates eq .
( [ eqn : mc ] ) to eq .
( [ generalestimator ] ) .
so , @xmath138 .
therefore , when neglecting the derivatives of @xmath135 , eq .
( [ generalestimator ] ) needs to be the starting point of the derivation . a number of non - trivial corrections to the estimator proposed by result from this starting point . to simplify the expressions that follow , using the notation of we define the ` schur ' parameter @xmath139 as @xmath140 where the coefficents of @xmath119 are defined in eq .
( [ eqn : a ] ) .
additionally , we define @xmath141 as the cross - correlation coefficient between the photo-@xmath3 sample and the @xmath72 redshift bin of the spec-@xmath3 sample : @xmath142 . adopting these definitions
, the full estimator can be written as @xmath143_{\rm last } + \sum_j ( f^{-1})_{ij } \sum_{\ell , m } \ , \left ( \frac{s}{a_{00 } \ , a_{jj } }
\right ) \frac{\partial a_{0j}}{\partial p_j } \nonumber \\ & \bigg [ & \sum_k \left ( \delta^k_{jk } + 2 s r_j r_k \sqrt{\frac{a_{jj}}{a_{kk } } } \right ) ( \widehat{p } \ ; \widehat{s}_k - a_{0k } ) \nonumber \\ & - & \sum_k \frac{a_{0k}}{a_{kk } } \left ( \delta^k_{jk } + s r_j r_k \sqrt{\frac{a_{jj}}{a_{kk } } } \right ) ( \widehat{s}_k \ ; \widehat{s}_k - a_{kk } ) \nonumber \\ & - & \frac{s a_{0j}}{a_{00 } } ( \widehat{p } \ ; \widehat{p } - a_{00 } ) \bigg ] \ , , \label{estimatorfull } \end{aligned}\ ] ] where @xmath144 is the kronecker delta . in this expression
the ( @xmath145 ) dependence of the multipole coefficients is implicit , so , @xmath146 and @xmath147 .
one can check this expression converges to the input theory as follows .
first , we write the correction term as @xmath148_{\rm last}$ ] . now , assuming the input theory is correct , @xmath149 , @xmath150 , and @xmath151 .
then , following some algebra , one can show eq .
[ estimatorfull ] implies @xmath152 , thus , proving that the estimator will converge . for the equivalent equation from
, one can show @xmath153 .
the fisher matrix ( eq . [ eqn : fish1 ] ) remains unchanged from , where @xmath154_{,i } \ ; [ a_{0j}]_{,j } \ , .
\label{eqn : limber_fish}\ ] ] in the limit where shot noise dominates ( i.e. , where @xmath155 and @xmath156 ) and neglecting auto - correlations , our result ( eq . [ estimatorfull ] ) agrees with eq .
( 36 ) from .
when analyzing observational data we will work exclusively in configuration space , which allows us to avoid difficulties with complex survey geometries ( in the future this may not be necessary ( see , * ? ? ?
* ; * ? ? ?
thus , in order to match our theory with observations , we need to convert our estimator in eq .
( [ estimatorfull ] ) from harmonic to configuration space .
this conversion is simplified by the following relation : @xmath157 \nonumber \\ & = 8 \pi^2 \int dx \ , v_i(x ) \ , \widehat{w}_{p s_i}(x ) \nonumber \\ & \approx 8 \pi^2 \sum_{\alpha } \delta \theta_{\alpha } \theta_{\alpha } \ , v_i(\theta_{\alpha } ) \
, \widehat{w}_{p s_i}(\theta_{\alpha } ) \ , , \label{eqn : converttheta}\end{aligned}\ ] ] where @xmath158 is the observed angular cross - correlation function , @xmath159 is the angular separation scale , and @xmath160 . here
we have explicitly subtracted the shot - noise component .
( [ eqn : converttheta ] ) is valid for an arbitrary function @xmath161 , which is related to @xmath162 by @xmath163 where @xmath164 are the legendre polynomials .
our measurements of the angular correlation functions are made in bins of width @xmath165 , with central values @xmath166 .
these values set the properties of the summation in eq .
( [ eqn : converttheta ] ) . note , because the kernel @xmath167 is not a slowly varying function , a narrow @xmath159 spacing @xmath168 is needed to accurately approximate this integral .
now , to convert eq .
( [ estimatorfull ] ) into configuration space we first re - write the estimator in terms of four weighting functions defined as @xmath169 and @xmath170 finally , using eq . ( [ eqn : converttheta ] ) one finds that our estimator in configuration space takes the form @xmath171_{\rm last } + 8 \pi^2 \sum_j ( f^{-1})_{ij } \label{eqn : fullconfig } \\ & \sum_{\alpha } \delta \theta_{\alpha } \ , \theta_{\alpha}\bigg [ \nonumber d_j(\theta_{\alpha } ) \,\bigg\ { \widehat{w}_{ps_j}(\theta_{\alpha } ) - w_{ps_j}(\theta_{\alpha } ) \bigg\ } \nonumber \\ & + h_j(\theta_{\alpha } ) \ , \widehat{w}_{s_js_j}(\theta_{\alpha})- g_j(\theta_{\alpha } ) \ , \widehat{w}_{pp}(\theta_{\alpha } ) \nonumber \\ & + \sum_{k } e_{jk}(\theta_{\alpha})\bigg\ { \widehat{w}_{p s_{k}}(\theta_{\alpha } ) - w_{ps_k}(\theta_{\alpha } ) + \frac{1}{2}\widehat{w}_{s_ks_k}(\theta_{\alpha } ) \bigg\ } \bigg ] \nonumber \ , , \end{aligned}\ ] ] where @xmath172 have been transformed to configuration space using eq .
( [ eqn : converttheta2 ] ) . ignoring both the bin - to - bin correlations , such that the fisher matrix is diagonal , and the auto - correlation terms , the second term in eq .
( [ eqn : fullconfig ] ) becomes @xmath173 \ , .\end{aligned}\ ] ]
the result is now much more intuitive .
the @xmath92 is reconstructed from a weighted minimization of @xmath174 , where the weights are given by @xmath175 .
note that the scales that contribute most to the estimator are represented by the combination @xmath176 . to illustrate the angular sensitivity of the estimator , in fig .
( [ weights ] ) we plot these weights for the mocks introduced in the next section . we find that the weights peak at @xmath177 arcmins .
however , as emphasized by , the breadth and smoothness of the weighting function implies that the fourier modes being traced are on quasi - linear scales : sharp cuts in angle have a greater sensitivity to non - linearity than a smoother filter .
in this section , we test our methodology using 20 sets of mock halo catalogues mock catalogues is sufficient for the level of error we wish to investigate . ]
created from the n - body simulations described in sec .
( [ sec : simulations ] ) . for each 60 deg@xmath27 simulation we generated a uniform redshift distribution of mock spectroscopic sources within the range @xmath178 , adopting an angular density of 1000 sources deg@xmath179 .
in addition , we sampled mock photometric galaxies using a gaussian redshift distribution with mean 0.5 and standard deviation 0.1 with a density of 1 source arcmin@xmath179 ( which roughly mimicks a typical tomographic bin in kids ) .
for the purposes of this test we generated each sample by randomly sampling from the halo catalogue at each redshift , such that the bias factors of the photometric and spectroscopic samples are expected to be the same ( this would not necessarily be true for a real data sample ) .
we then performed a cross - correlation analysis dividing the spectroscopic sources into 16 redshift bins of width @xmath180 .
for further details on the mock catalogues we refer the reader to @xcite , section 6.1 .
we note that each individual mock provides constraints comparable in precision to the observational datasets used in our analysis ( each mock realization contains @xmath181 spec-@xmath3 galaxies , compared to @xmath22 spec-@xmath3 galaxies in 2dflens ) .
we measured the angular auto- and cross - correlation functions in our analysis using the landy - szalay estimator @xcite , generating random catalogues 10 times larger than the data distribution .
for example , the cross - correlation between samples @xmath21 and @xmath57 is @xmath182 where @xmath183 , @xmath184 and @xmath185 are the respective pair counts between the two data samples , the data and random samples , and the two random samples , as a function of the angular separation @xmath159 .
the number counts of the data sample @xmath21 and the random sample @xmath57 are @xmath186 and @xmath187 , respectively .
the correlation functions are all measured using 30 equally logarithmically - spaced angular bins between 0.01 and 1 deg .
we estimate the errors in the measurements using jack - knife re - sampling , although these errors are not used in the quadratic estimation process . for each mock catalogue we measured the following statistics :
the auto - correlation of the photometric galaxies , @xmath188 ; the @xmath189 spectroscopic auto - correlations in each redshift bin , @xmath190 ; and the @xmath189 photometric - spectroscopic cross - correlations , @xmath191 , where the index @xmath21 runs across the @xmath189 redshift bins of the spectroscopic sample . in this section
we apply the quadratic estimator to the mock catalogues and infer results for the redshift probability distribution @xmath42 across the spectroscopic bins . for the purposes of this test we only use the angular cross - correlation functions as inputs to the quadratic estimator .
thus , we apply eq .
( [ eqn : fullconfig ] ) and drop the auto - correlation terms .
we note that in this limit the estimator remains unbiased , as the auto - correlation terms cancel .
as discussed in the next section , we use the auto - correlations of the spec-@xmath3 sample separately , to measure the redshift evolution of the galaxy bias .
every iteration of the quadratic estimator returns a correction term to the probability distribution @xmath192 computed from the previous best - guess estimate of the redshift distribution .
we define the estimator to be _ converged _ once the condition @xmath193 is met . at this level of accuracy ,
the estimated correction is on average an order of magnitude smaller than the error for a given redshift bin , i.e. , @xmath194 . after each iteration of the quadratic estimator
, we also enforce the normalization condition @xmath195 .
this produces an overall amplitude shift that is minimal in most cases , although we do expect a small bias to be introduced when imposing this constraint because the normalization condition holds only for the underlying @xmath196 , not the estimated @xmath197 .
we can assess the convergence of the estimator and the accuracy of the modelling by comparing the observed , reconstructed and model angular cross - correlation functions obtained from a single ( representative ) mock catalogue , as shown in fig .
( [ crosscorr_test ] ) .
each panel within this figure illustrates the cross - correlation function @xmath191 for one of the spectroscopic redshift bins , starting from the lowest redshift in the top left - hand corner .
we overplot the model prediction for the angular cross - correlations as the red lines , derived using eq .
( [ wtheta2 ] ) and using the bias of the samples determined as discussed in sec .
( [ subsec : bias ] ) below .
the blue lines in fig .
( [ crosscorr_test ] ) show the _ reconstructed _ angular cross - correlation functions , which we obtain by using the recovered photo-@xmath3 redshift distribution , @xmath198 , to compute the angular cross - correlation via @xmath199 we can now assess the convergence of the estimator by comparing the reconstructed predictions ( blue lines ) to the mock measurements ( blue points ) .
thus , from fig .
( [ crosscorr_test ] ) , we observe that the combination @xmath174 is being successfully minimized .
one should keep in mind the effective @xmath159-dependent weights being applied to this minimization , as shown by fig .
( [ weights ] ) . comparing cross - correlation functions
is a useful validation of the estimator , as this test is less sensitive to inaccuracies in modelling the correlation functions and galaxy bias than comparing the inferred @xmath200 distribution to the input relative to the true value .
however , in this case @xmath201 should still agree with the simulation prediction . ] .
systematic errors in the modelling of the cross - correlation statistics for example , due to the breakdown of an assumption such as linear galaxy bias could propagate into a bias in the inferred photometric redshift distribution .
we assessed this potential source of systematic error by comparing the average of the mock cross - correlation function measurements to our model predictions , the results of which are shown in fig .
( [ model_test ] ) .
the models are generated using the galaxy bias determined for the spectroscopic sample in sec .
( [ subsec : bias ] ) below , and in this figure we plot the average error for individual mocks .
we conclude that our modelling of the angular cross - correlation function is sufficient at the level of statistical errors present in a single mock catalogue .
however , if we instead use the error in the mock mean ( dividing by @xmath202 ) we observe some tension between the simulation predictions and the analytic modelling of the correlation functions .
these tests reveal that there are non - linear effects in the mocks not captured by our model ( e.g. , non - linear galaxy bias ) , and also that the jack - knife errors do not fully capture the scatter in the measurements .
these issues could be mitigated by restricting our analyses to larger scales , and by deducing the statistical errors using a dispersion in the mocks , rather than by jack - knife techniques .
however , we leave such investigations to future work . in order to test
whether our quadratic estimation pipeline recovers the input source redshift distribution of the mocks , we also require the redshift evolution of the galaxy bias factors , which we have arranged ( by sampling halos in the same mass range ) to be the same for the spectroscopic and photometric samples .
we determined the redshift evolution of this bias using the auto - correlation function measurements of each spec-@xmath3 sample , using chi - squared minimization to fit for @xmath203 .
we note that the effects of noise will cause fluctuations between @xmath203 and @xmath204 . in particular , given the significantly lower number density of spec-@xmath3 galaxies relative to the photo-@xmath3 sample ,
the bias measurements of the spec-@xmath3 sample will be less accurate .
we overcome this by averaging the @xmath203 values over the 20 mocks .
we propagate the noise in the measurement of @xmath203 into the inferred redshift probability distributions by empirically determining that the scatter in the bias measurements across mocks , @xmath205 , produces a scatter in the probability distribution @xmath206 .
we now consider the results for the recovered redshift distributions of the mock source catalogues , measured in 16 step - wise redshift bins of width @xmath30 .
we present a random subset of these results for 9 mocks in fig .
( [ mock_results ] ) .
the green data points display the best - fitting @xmath207 values and the 1-@xmath208 errors are derived from the fisher matrix .
the black line shows the gaussian redshift distribution assumed in the simulations .
we can see that the reconstruction has proved generally successful . for applications to weak gravitational lensing ,
a flexible functional form for the redshift distribution is more convenient than a step - wise binning .
we adopt _ gaussian processes _ ( gps ) as a method to infer such functional forms for redshift distributions .
briefly , gps provide _ non - parametric _ bayesian modelling for supervised learning problems . for details
, we refer the reader to @xcite .
gps also readily allow for the inclusion of a prior on the smoothness of the reconstructed function , naturally expected from a survey selection function
. will necessarily be a very smooth function of redshift , as photo-@xmath3 cuts can cause sharp variations .
rather , we are implying that the amplitude between redshift bins will be highly correlated .
thus , this setup still allows sharp fluctuations , although they are less probable . ]
we build gp models using the python module scikit - learn @xcite . as input one needs to define the functional form for the adopted ( redshift ) correlation function @xmath209 and set the _ characteristic scale _ @xmath210 , which intuitively determines the typical distance between peaks ( i.e. , the smoothness scale ) of the function .
we set the redshift correlation as a gaussian function @xmath211 \ , , \ ] ] where @xmath212 defines the variance of the function , and we set @xmath213
. the optimal choice of @xmath210 and @xmath214 will depend on the survey in question . in fig .
( [ mock_results ] ) we display the 2-@xmath208 confidence intervals for the gp models as the green band .
we observe that the reconstructed gp distributions encompass the input mock redshift distribution for the vast majority of mocks and redshift bins .
more quantitatively , using all 20 mock catalogues we measure an average @xmath215 of @xmath216 for 18 degrees of freedom ( ignoring bin - to - bin correlations ) demonstrating that , at the level of statistical error of _ individual _ mock catalogues ( and hence of the observational datasets used in our study ) , our measurements are statistically consistent with the input distribution .
we now consider a more accurate test of our methodology , using the average @xmath42 values over all 20 mock catalogues . for this test , we consider two methods to estimate the bias of the photo-@xmath3 sample , the results of which are shown in fig .
( [ mock_results_ave ] ) . for _ method 1
_ we adopt our default model , i.e. , @xmath217 , where we average the spec-@xmath3 bias factors over 20 mocks ( blue points and band ) . for _ method 2 _ we use a bias evolution model with two free parameters defined as @xmath218 , where we have fitted @xmath219 and @xmath220 from the mocks ( green points and band ) .
the input gaussian redshift distribution is displayed as the black line . inspecting fig .
( [ mock_results_ave ] ) , we find a significant discrepancy between our predictions based on method 1 and the input distribution .
qualitatively , we observe that for @xmath221 the distribution tends to be underestimated , while for @xmath222 the distribution tends to be overestimated .
we interpret these discrepancies as a limitation of our modelling of the cross - correlation function and the influence of non - linear galaxy bias .
we find that the predictions using the second bias model are more representative .
in this section , we apply the quadratic estimation methodology to infer the product @xmath223 for sources detected in the separate @xmath15-band and @xmath21-band catalogues of the kids imaging survey , using their cross - correlation with the 2dflens spec-@xmath3 catalogue . because ( unlike for the mock catalogues ) we have no information on @xmath65 , we can not break the degeneracy @xmath223 without further assumptions .
we cross - correlate the photometric samples with the 2dflens catalogue in 18 redshift bins in the range @xmath31 .
the @xmath15-band and @xmath21-band samples have a different degree of overlap with 2dflens , and we use a total of @xmath224 and @xmath225 spec-@xmath3 galaxies for the respective cross - correlations . following the cosmic shear analysis of @xcite , we divided the kids-450 @xmath15-band imaging dataset into four tomographic bins based on the bpz redshift @xmath226 : @xmath227 , @xmath228 , @xmath229 and @xmath230 .
photo-@xmath3 information is not available for the kids-800 @xmath21-band dataset , and we do not sub - divide it . in the correlation function measurement , we now weight each source by its optimal weight in the estimation of shear statistics , such that we are constraining the weighted redshift probability distribution of the sources @xcite . since we are only utilizing spectroscopic data in the range @xmath23 , we can not derive the full kids source redshift distributions for most of the samples .
rather , our motivation is to demonstrate an application of our methodology on a realistic dataset .
we fit for the spec-@xmath3 bias values in each redshift bin as outlined in the previous section , and show the results in fig .
( [ bias_evolv ] ) .
we perform a rough scaling of the expected error in the bias compared to the mocks , and propagate this error into the determination of @xmath231 by adding a term @xmath232 to the standard fisher matrix errors .
it is possible for measurements of the angular cross - correlation function to be negative due to either noise , or effects such as incompleteness in the imaging catalogue around bright spectroscopic sources or other systematics .
these points are unphysical in our model , which predicts the cross - correlation functions by scaling the auto - correlations , which are positive definite . to address this problem we add a positive definite prior which effectively shifts negative @xmath231 values to zero . in fig .
( [ rband_results ] ) we show the reconstructed @xmath233 measurements for each tomographic bin of the kids-450 @xmath15-band data , in comparison with other determinations of this distribution presented by @xcite .
the green data points show the quadratic estimation with 1-@xmath208 error bars , and the shaded green band gives the @xmath234 confidence interval for the gp model .
only 14 redshift bins are shown per tomographic bin ; it was not possible to recover an estimate for the first two and last two redshift bins because of the low number of spec-@xmath3 galaxies in these bins , which caused instabilities in the estimator , hence poor convergence .
the red and blue bands display the distributions and @xmath234 confidence ranges obtained by applying two other methods : * calibration with small - scale cross - correlations ( cc ) , shown by the red bands and implemented by applying the recipes of @xcite and @xcite to the cross - clustering of kids sources and deep spectroscopic samples from deep2 and cosmos ; * weighted direct calibration ( dir ) , shown by the blue bands and based on direct determination of the source redshift distribution by cross - matching with a number of overlapping spectroscopic samples , with appropriate re - weighting for incompleteness .
we refer the reader to @xcite for more details about these methods .
we converted the dir estimation from @xmath42 to @xmath235 using the function @xmath65 implicitly assumed in the cc method , as outlined by @xcite .
for the purposes of this comparison , all the distributions have been normalized such that @xmath236 .
these different methods produce redshift distributions which agree in a qualitative sense , although comparisons illustrate systematic errors affecting each technique . the equivalent quadratic estimation for the kids-800 @xmath21-band imaging data , analyzed in a single tomographic bin , is shown in fig .
( [ iband_results ] ) . in this figure
the blue points display the reconstructed redshift distribution with 1-@xmath208 errors , and the shaded blue region shows the @xmath237 confidence interval for the gp model .
we observe that the amplitude of the cross - correlation signal increases with redshift , driven by a combination of the underlying redshift distribution @xmath42 and the source galaxy bias @xmath65 . assuming that the latter is a slowly - varying function , this analysis suggests that the redshift distribution of the @xmath21-band sources is broad and peaked at @xmath238 .
these findings are qualitatively consistent with the dir estimate presented by , although further comparison is beyond the scope of the current study .
these results demonstrate that , although the 2dflens dataset does not have sufficient redshift coverage to derive the full source distributions by cross - correlation , we can successfully apply our methodology to the kids dataset and obtain results in qualitative agreement with previous determinations .
with the issue of source redshift calibration becoming increasingly pronounced for weak gravitational lensing surveys , new and versatile approaches to this problem are needed .
calibration via cross - correlation with overlapping spectroscopic surveys represents one such approach . in this work ,
we have presented our efforts to extend the accuracy and applicability of such methods to both simulations and data .
we summarise our main findings as follows : * we have developed a new , minimum - variance and unbiased _ quadratic estimator _ that infers the redshift probability distributions of photometric samples of galaxies @xmath42 , in a degenerate combination with their galaxy bias @xmath65 , based on their angular cross - correlation with an overlapping spectroscopic sample .
this derivation expands on work presented by @xcite .
* we have tested our methodology on a series of mock galaxy catalogues .
we found that at the level of statistical errors of current surveys the estimator is unbiased .
however , if we stack mocks together to perform a more accurate test , we discover small but significant discrepancies we attribute these effects to the breakdown of our clustering model due , for example , to non - linear galaxy bias . *
we derive non - parametric , continuous functional forms of @xmath239 by building _ gaussian process _
models from the step - wise constraints inferred from the quadratic estimator .
such continuous functions are useful for modelling the lensing signal , and allow the computation of continuous confidence intervals .
* we have applied our methodology to infer @xmath239 functions for kids @xmath15-band and @xmath21-band imaging catalogues in the range @xmath240 , through cross - correlation with the 2dflens spectroscopic redshift survey .
our distributions are in qualitative agreement with the results of other methods .
our analysis could be extended in a number of ways : by improving the modelling of non - linear effects , enhancing our mock catalogues to match our datasets more closely , and re - formulating the estimator to measure the spectroscopic galaxy bias and redshift distribution simultaneously .
however , the calibration of the bias of the photometric sample remains the most critical component .
possible approaches to this problem include the use of redshift - space distortions , lensing magnification , and galaxy - galaxy lensing .
we hope that our work motivates more research on these topics .
we thank matt mcquinn and martin white for very helpful discussions . the 2dflens survey is based on data acquired through the australian astronomical observatory , under program a/2014b/008 .
it would not have been possible without the dedicated work of the staff of the aao in the development and support of the 2df - aaomega system , and the running of the aat .
this study is based on data products from observations made with eso telescopes at the la silla paranal observatory under programme ids 177.a-3016 , 177.a-3017 and 177.a-3018 , and on data products produced by target / omegacen , inaf - oacn , inaf - oapd and the kids production team , on behalf of the kids consortium .
omegacen and the kids production team acknowledge support by nova and nwo - m grants .
members of inaf - oapd and inaf - oacn also acknowledge the support from the department of physics & astronomy of the university of padova , and of the department of physics of univ .
federico ii ( naples ) .
computations for the @xmath93-body simulations were performed in part on the orcinus supercomputer at the westgrid hpc consortium ( www.westgrid.ca ) , in part on the gpc supercomputer at the scinet hpc consortium .
scinet is funded by : the canada foundation for innovation under the auspices of compute canada ; the government of ontario ; ontario research fund - research excellence ; and the university of toronto . | we develop a statistical estimator to infer the redshift probability distribution of a photometric sample of galaxies from its angular cross - correlation in redshift bins with an overlapping spectroscopic sample .
this estimator is a minimum - variance weighted quadratic function of the data : a _
quadratic estimator_. this extends and modifies the methodology presented by @xcite .
the derived source redshift distribution is degenerate with the source galaxy bias , which must be constrained via additional assumptions .
we apply this estimator to constrain source galaxy redshift distributions in the kilo - degree imaging survey through cross - correlation with the spectroscopic 2-degree field lensing survey , presenting results first as a binned step - wise distribution in the range @xmath0 , and then building a continuous distribution using a gaussian process model .
we demonstrate the robustness of our methodology using mock catalogues constructed from n - body simulations , and comparisons with other techniques for inferring the redshift distribution
. surveys , cosmology : observation , large scale structure of the universe |
when a robotic vehicle equipped with a vision sensor is manoeuvering in a built environment , consisting primarily of planar or near planar surfaces , then the nature of the environment can be exploited in the vision processing algorithms .
different images of the same planar surface are related by homography mappings , and homographies have been used extensively in robotic applications as a vision primitive .
homography - based algorithms have been used for estimation of the rigid - body pose of a vehicle equipped with a camera @xcite .
navigation of robotic vehicles has been developed based on homography sequences @xcite and one of the most successful visual servo control paradigms uses homographies @xcite .
homography - based methods are particularly well suited to navigation of unmanned aerial vehicles @xcite where the ground terrain is viewed from a distance and the relief of surface features is negligible compared to the vehicles distance from the scene .
computing homographies from point correspondences has been extensively studied in the last fifteen years @xcite .
the quality of the homography estimate obtained depends heavily on the number and quality of the data points used in the estimation as well as the algorithm employed . for a well textured scene
, the state - of - the - art methods can provide high quality homography estimates at the cost of significant computational effort ( see @xcite and references therein ) .
for a scene with poor texture , and consequently few reliable feature point correspondences , existing homography estimation algorithms perform poorly .
robotic vehicle applications , however , provide temporal sequences of images and it seems natural to exploit the temporal correlation rather than try to compute individual raw homographies for each pair of frames .
zhang @xcite used image flow computed from a pair of images to compute the relative homography , although this method still only considers isolated pairs of images . in recent work by the authors @xcite a nonlinear observer @xcite for homography estimation was proposed based on the group structure of the set of homographies , the special linear group @xmath0 @xcite .
this observer uses velocity information to interpolate across a sequence of images and improve the individual homography estimates .
the observer proposed in @xcite still requires individual image homographies to be computed for each image , which are then smoothed using filtering techniques .
although this approach @xcite provides improved homography estimates , it comes at the cost of running both a classical homography algorithm as well as the temporal filter algorithm , and only functions if each pair of images has sufficient data available to compute a raw homography . in this paper
, we consider the question of deriving an observer for a sequence of image homographies that takes image point feature correspondences directly as input .
the proposed approach takes a sequence of images associated with a continuous variation of the reference image , the most common case being a moving camera viewing a planar scene , a situation typical of robotic applications .
the proposed nonlinear observer is posed on the special linear group @xmath0 , that is in one - to - one correspondence with the group of homographies @xcite , and uses velocity measurements to propagate the homography estimate and fuse this with new data as it becomes available @xcite .
a key advance on prior work by the authors is the formulation of a point feature innovation for the observer that incorporates point correspondences directly in the observer without requiring reconstruction of individual image homographies .
this saves considerable computational resources and makes the proposed algorithm suitable for embedded systems with simple point tracking software .
in addition , the algorithm is well posed even when there is insufficient data for full reconstruction of a homography .
for example , if the number of corresponding points between two images drops below four it is impossible to algebraically reconstruct an image homography and the existing algorithms fail @xcite . in such situations , the proposed observer will continue to operate , incorporating what information is available and relying on propagation of prior estimates where necessary . finally , even if a homography can be reconstructed from a small set of feature correspondences , the estimate is often unreliable and the associated error is difficult to characterize .
the proposed algorithm integrates information from a sequence of images , and noise in the individual feature correspondences is filtered through the natural low - pass response of the observer , resulting in a highly robust estimate . as a result , the authors believe that the proposed observer is ideally suited for estimation of homographies based on small windows of image data associated with specific planar objects in a scene , poorly textured scenes , and real - time implementation ; all of which are characteristic of requirements for homography estimation in robotic vehicle applications . the paper is organized into seven sections including the introduction and the concluding sections .
section [ sec : background ] presents a brief recap of the lie group structure of the set of homographies . in section [ sec : observer ] , based on a recent advanced theory for nonlinear observer design @xcite , a nonlinear observer on @xmath1 is proposed using direct measurements of known inertial directions and the knowledge of the group velocity .
a rigourous stability analysis is provided in this section . in section
[ sec : application ] , the homography and associated homography velocity are related to rigid - body motion of the camera and two observers are derived for the case where only the angular velocity of the camera is known , a typical scenario in robotic applications .
an initial constructed example in section [ section : experiment ] confirms the robust performance of the proposed algorithm on real image data .
section [ sec : experimentii ] provides an application of our approach to a real world problem in image stabilization .
two videos , showing the results of sections [ section : experiment ] and [ sec : experimentii ] , respectively , are provided as supplementary material .
the results show excellent performance even in the case of very fast camera motion ( relative to frame rate ) , severe occlusion , and in the presence of specular reflections .
visual data is obtained via a projection of an observed scene onto the camera image surface .
the projection is parameterised by two sets of parameters : intrinsic ( `` internal '' parameters of the camera such as the focal length , the pixel aspect ratio , etc . ) and extrinsic ( the pose , i.e. the position and orientation of the camera ) .
let @xmath2 ( resp .
@xmath3 ) denote projective coordinates for the image plane of a camera @xmath4 ( resp .
@xmath5 ) , and let @xmath6 ( resp .
@xmath7 ) denote its ( right - hand ) frame of reference .
let @xmath8 denote the position of the frame @xmath7 with respect to @xmath6 expressed in @xmath6 .
the orientation of the frame @xmath7 with respect to @xmath6 , is given by a rotation matrix @xmath9 , where @xmath10 as a mapping .
the pose of the camera determines a rigid body transformation from @xmath7 to @xmath6 ( and visa versa ) .
one has @xmath11 as a relation between the coordinates of the same point in the reference frame ( @xmath12 ) and in the current frame ( @xmath13 ) .
the camera internal parameters , in the commonly used approximation @xcite , define a @xmath14 matrix @xmath15 so that we can write . ] : @xmath16 where @xmath17 is the image of a point when the camera is aligned with frame @xmath7 , and can be written as @xmath18 using the homogeneous coordinate representation for that 2d image point . likewise , @xmath19 is the image of the same point viewed when the camera is aligned with frame @xmath6 .
if the camera is calibrated ( the intrinsic parameters are known ) , then all quantities can be appropriately scaled and the equation is written in a simple form : @xmath20 since homographies describe image transformations of _ planar _ scenes , we begin by fixing a plane that contains the points of interest ( target points ) . assume a calibrated camera and that there is a planar surface @xmath21 containing a set of @xmath22 target points ( @xmath23 ) so that @xmath24 where @xmath25 is the unit normal to the plane expressed in @xmath6 and @xmath26 is the distance of the plane to the origin of @xmath27 . from the rigid - body relationships , one has @xmath28 .
define @xmath29 .
since all target points lie in a single planar surface one has @xmath30 and thus , using , the projected points obey @xmath31 the projective mapping @xmath32 , @xmath33 is termed a homography and it relates the images of points on the plane @xmath34 when viewed from two poses defined by the coordinate systems @xmath3 and @xmath35 , respectively .
it is straightforward to verify that the homography @xmath36 can be written as follows : @xmath37 where @xmath38 is the normal to the observed planar surface expressed in the frame @xmath7 and @xmath39 is the orthogonal distance of the plane to the origin of @xmath7 .
one can verify that @xcite : @xmath40 the homography matrix contains the pose information @xmath41 of the camera from the frame @xmath7 ( termed current frame ) to the frame @xmath6 ( termed reference frame ) .
however , since the relationship between the image points and the homography is a projective relationship , it is only possible to determine @xmath36 up to a scale factor ( using the image points relationships alone ) .
recall that the special linear group @xmath0 is defined as the set of all real valued @xmath42 matrices with unit determinant : @xmath43 the lie - algebra @xmath44 for @xmath0 is the set of matrices with trace equal to zero : @xmath45 .
the adjoint operator is a mapping @xmath46 defined by : @xmath47 since a homography matrix @xmath36 is only defined up to scale then any homography matrix is associated with a unique matrix @xmath48 by re - scaling @xmath49 such that @xmath50 .
every such matrix @xmath48 occurs as a homography .
moreover , the map @xmath51 is a group action of @xmath0 on the projective space @xmath52 since @xmath53 where @xmath54 and @xmath55 and @xmath56 is the identity matrix , the unit element of @xmath0 .
the geometrical meaning of the above property is that the 3d motion of the camera between views @xmath57 and @xmath58 , followed by the 3d motion between views @xmath58 and @xmath59 is the same as the 3d motion between views @xmath57 and @xmath59 . as a consequence
, we can think of homographies as described by elements of @xmath0 . since any homography is defined up to a scale factor , we assume from now on that @xmath60 : @xmath61 there are numerous approaches for determining @xmath36 , up to this scale factor , cf .
for example @xcite .
note that direct computation of the determinant of @xmath36 in combination with the expression of @xmath39 and using the fact that @xmath62 , shows that @xmath63 .
extracting @xmath64 and @xmath65 from @xmath36 is in general quite complex @xcite and is beyond the scope of this paper .
in this section , the design of an observer for @xmath66 is based on a recent theory for nonlinear observer design directly on the output space @xcite .
consider the kinematics of @xmath1 given by @xmath67 with @xmath68 the group velocity .
assume that @xmath69 is measured .
furthermore , we dispose of a set of @xmath22 measurements @xmath70 in the body - fixed frame : @xmath71 where @xmath72 are constant and known . for later use , define @xmath73 a set @xmath74 of @xmath75 vector directions @xmath72 , with @xmath76 , is called _ consistent _ , if it contains a subset @xmath77 of @xmath78 constant vector directions such that all vector triplets in @xmath79 are linearly independent .
this definition implies that if the set @xmath80 is consistent then , for all @xmath81 there exists a unique set of three non vanishing scalars @xmath82 ( @xmath83 ) such that @xmath84 we verify that @xmath1 is a symmetry group with group actions @xmath85 , @xmath86 and @xmath87 defined by @xmath88 indeed , it is straightforward to verify that @xmath89 , @xmath90 , and @xmath91 are _ right group actions _ in the sense that @xmath92 , @xmath93 , and @xmath94 , for all @xmath95 , @xmath96 , and @xmath97 . clearly , @xmath98 and @xmath99 & = & hu q = ( hq ) ( q^{-1 } uq ) \\ & = & f(\phi(q , h ) , \psi(q , u ) ) .
\end{array}\ ] ] thus , the kinematics are _ right equivariant _ ( see @xcite ) .
let @xmath100 denote the estimate of @xmath36 .
define the right group error @xmath101 and the output errors @xmath102 : @xmath103 inspired by @xcite , the proposed observer takes the form @xmath104 where @xmath105 is the innovation term to be designed and must be _ right equivariant _ in the sense that for all @xmath106 : @xmath107 interestingly , if the innovation term @xmath108 is equivariant , the dynamics of the canonical error @xmath109 is autonomous and given by ( * ? ? ?
1 ) : @xmath110 in order to determine @xmath111 , we need a _ non - degenerate right - invariant _ cost function . to this purpose ,
let us first define individual _ degenerate right - invariant costs _ at @xmath112 on the output space @xmath52 as follows : @xmath113 with @xmath114 positive numbers .
one verifies that @xmath115 are right - invariant in the sense that @xmath116 for all @xmath106 .
the costs @xmath115 are degenerate since by taking @xmath117 there exists an infinity of @xmath118 such that @xmath119 .
then , the aggregate cost is defined as the sum of all the individual costs as follows : @xmath120 it is straightforward that @xmath121 is right - invariant . according to ( * ? ? ? * lem .
3 ) , the aggregate cost is _ non - degenerate _ if @xmath122 where the stabilizer @xmath123 of an element @xmath97 is defined by @xmath124 .
in fact , @xmath125 is equivalent to @xmath126 , where @xmath127 is the lie - algebra associated with @xmath128 ( see @xcite ) . [ : lem : aggregatecost ] assume that the set @xmath74 of the measured directions @xmath112 is consistent .
then , the aggregate cost @xmath121 defined by is non - degenerate . as a consequence ,
@xmath129 is a global minimum of the aggregate cost @xmath130 .
one computes the derivative @xmath131 & \!\!\!\!=\!\!\!\ ! & \mathrm{d } \left(\frac{h^{-1}\mathring{p}_i}{|h^{-1}\mathring{p}_i|}\right ) [ hu]\\ & \!\!\!\!=\!\!\!\ ! & \left(i - \frac{(h^{-1}\mathring{p}_i)(h^{-1}\mathring{p}_i)^\top}{|h^{-1}\mathring{p}_i|^2 } \right ) \frac{uh^{-1}\mathring{p}_i}{|h^{-1}\mathring{p}_i| } \end{array}\ ] ] with some @xmath132 . from
, one deduces that @xmath133 with @xmath134 for all @xmath135 .
thus , @xmath136 now , let us determine @xmath132 such that @xmath137 , for all @xmath138 .
the relation @xmath137 can be equivalently written as @xmath139 with @xmath140 . from here , one deduces that @xmath141 are eigenvalues of @xmath69 and @xmath142 are the associated eigenvectors .
since the set @xmath74 of the measured directions @xmath112 is consistent , without loss of generality we assume that the subset @xmath143 is consistent .
thus , there exist 3 non - null numbers @xmath144 , @xmath145 , and @xmath146 such that @xmath147 . from here , using the fact that @xmath148 and @xmath149 one deduces @xmath150 since @xmath151 ( with @xmath152 ) are non - null and the 3 unit vectors @xmath112 ( with @xmath152 ) are linearly independent , directly yields @xmath153 .
let @xmath154 denote the value of these four identical eigenvalues .
from here , one easily deduces that the _ geometric multiplicity _ of the eigenvalue @xmath154 ( defined as the dimension of the eigenspace associated with @xmath154 ) is equal to 3 , since the 3 eigenvectors @xmath112 ( with @xmath152 ) associated with @xmath154 are linearly independent . since the _ algebraic multiplicity _ of the eigenvalue @xmath154 is no less than the corresponding _ geometric multiplicity _
, one deduces that it is also equal to 3 .
this means that @xmath69 has a triple eigenvalue @xmath154 .
since the number of linearly independent eigenvectors of @xmath69 is equal to 3 , the matrix @xmath69 is diagonalizable .
then , the diagonalizability of @xmath69 along with the fact that it has a triple eigenvalue implies that @xmath155 .
this in turn yields @xmath156 , which is null since @xmath69 is an element of @xmath157 .
consequently , @xmath158 and @xmath159 .
one then deduces @xmath126 which concludes the proof .
now that the non - degenerate right - invariant cost function @xmath121 is defined , we compute the innovation term @xmath111 as ( * ? ? ?
* eq . ( 40 ) ) @xmath160 where @xmath161 is the _ gradient _ in the first variable , using a _ right - invariant riemannian metric _ on @xmath0 .
let @xmath162 be a positive definite inner product on @xmath163 , chosen to be the euclidean matrix inner product on @xmath164 .
then , a right - invariant riemannian metric on @xmath0 induced by the inner product @xmath165 is defined by @xmath166 [ : lem : innovation ] the innovation term @xmath111 defined by is right equivariant and explicitly given by @xmath167 the proof for @xmath111 to be equivariant is a direct result of @xcite .
now , using standard rules for transformations of riemannian gradients and the fact that the riemannian metric is right - invariant , one obtains @xmath168&=\langle\mathrm{grad}_1 { \mathcal{c}}_{\mathring{p}}(\hat h , p ) , u \hat h \rangle_h \\ & = \langle \mathrm{grad}_1 { \mathcal{c}}_{\mathring{p}}(\hat h , p ) \hat h^{-1 } \hat h , u\hat h \rangle_h\\ & = \langle \mathrm{grad}_1 { \mathcal{c}}_{\mathring{p}}(\hat h , p ) \hat h^{-1 } , u \rangle\\ & = \langle \delta(\hat h , p ) , u \rangle , \end{split}\ ] ] with some @xmath132 . besides , in view of one has @xmath169
= d_{\hat h}{\mathcal{c}}_{\mathring{p}}(\hat h , p)[u \hat h ] \\[1ex ] & \,\ , = \sum_{i=1}^n k_i \left(\frac{\hat h p_i}{|\hat h p_i| } - \mathring p_i\right)^{\!\!\top } \!\ ! \left(i - \frac{(\hat h p_i)(\hat h p_i)^\top}{|\hat h p_i|^2}\right ) \frac{(u\hat h)p_i}{|\hat h p_i|}\\[1ex ] & \,\ , = \sum_{i=1}^n k_i(e_i -\mathring p_i)^\top ( i -e_ie_i^\top ) u e_i\\[1ex ] & \,\ , = \mathrm{tr}\left(\sum_{i=1}^n k_i e_i ( e_i -\mathring p_i)^\top ( i -e_ie_i^\top ) u \right ) \\[1ex ] & \,\ , = -\mathrm{tr}\left(\sum_{i=1}^n k_i e_i \mathring p_i^\top \pi_{e_i } u \right ) \\[1ex ] & \,\ , = \left \langle -\sum_{i=1}^n k_i\pi_{e_i}\mathring p_i e_i^\top , u \right\rangle .
\end{array}\ ] ] finally , the explicit expression of @xmath111 given by is directly obtained from and .
one deduces from that @xmath170 and , consequently , from that @xmath171 [ theo:1 ] consider the kinematics and assume that the velocity group @xmath132 is known .
consider the nonlinear observer defined by , with the innovation term @xmath172 defined by .
then , if the set @xmath80 of the measured directions @xmath112 is consistent , there exists a basin of attraction @xmath173 containing @xmath56 such that for any initial conditions @xmath174 and @xmath175 such that @xmath176 then @xmath177 and @xmath178 . this theorem can be seen as a direct result of theorem 2 in @xcite , but it can also be proved using classical lyapunov theory .
the candidate lyapunov function under consideration is @xmath179 . using the consistency of the set @xmath80
, one can ensure that @xmath180 is locally a positive definite function of @xmath109 .
the time - derivative of @xmath180 along the error flow verifies @xmath181 with @xmath182 denoting the frobenius norm defined by @xmath183 for any real valued square matrix @xmath5 . from here , one ensures that @xmath109 is locally bounded .
moreover , by application of lasalle s theorem , one deduces that @xmath184 converges to zero . from the definitions of @xmath185 and @xmath186 ,
one deduces that @xmath187 computing the trace of @xmath188 , it follows : @xmath189 define @xmath190 and @xmath191 , and it is straightforward to verify that @xmath192 using the fact that @xmath193 at the equilibrium and therefore @xmath194 , as well as the cauchy - schwarz inequality , it follows that @xmath195 and consequently one has : @xmath196 which in turn implies the existence of some scalars @xmath197 such that @xmath198 using and exploiting the consistency of the set @xmath80 , one can proceed analogously to the proof of lemma [ : lem : aggregatecost ] to deduce that @xmath109 has a triple eigenvalue @xmath154 and @xmath199 .
then , evoking the fact that @xmath200 , one deduces that @xmath201 and @xmath202 .
consequently , @xmath109 converges asymptotically to the identity @xmath56 .
[ rem : domainstab ] the boundaries of the stability domain associated with theorem [ theo:1 ] are extremely difficult , and probably impossible , to analytically characterise .
the nature of the lyapunov function @xmath203 is always well conditioned in the neighbourhood of the correct homography matrix , but the global geometry of @xmath0 is complex and there will be critical points and non - convex cost structure when the rotation component of the homography matrix has errors approaching @xmath204rads from the true homography .
the authors believe , based on extensive simulation studies ( cf .
[ section : experiment ] ) and our intuition for such problems , that the stability domain is very large in practice , encompassing all realistic scenarios where the camera actually observes the desired scene ( rotations of less than @xmath205rads and moderate displacements ) .
in this section , we consider the case where a sequence of homographies is generated by a moving camera viewing a stationary planar surface . the goal is to develop a nonlinear filter for the image homography sequence using the velocity associated with the rigid - body motion of the camera rather than the group velocity of the homography sequence , as was assumed in section [ sec : observer ] .
in fact , any group velocity ( infinitesimal variation of the homography ) must be associated with an instantaneous variation in measurement of the _ current _ image @xmath3 and not with a variation in the _ reference _ image @xmath2 .
this imposes constraints on two degrees of freedom in the homography velocity , namely those associated with variation of the normal to the reference image , and leaves the remaining six degrees of freedom in the homography group velocity depending on the rigid - body velocities of the camera .
denote the rigid - body angular velocity and linear velocity of @xmath7 with respect to @xmath6 expressed in @xmath7 by @xmath206 and @xmath207 , respectively .
the rigid body kinematics of @xmath41 are given by : @xmath208 where @xmath209 is the skew symmetric matrix associated with the vector cross - product , i.e. @xmath210 , for all @xmath211 .
recalling , it is easily verified that : @xmath212 this constraint on the variation of @xmath38 and @xmath26 is precisely the velocity constraint associated with the fact that the reference image is stationary . consider a camera attached to the moving frame @xmath3 moving with kinematics and viewing a stationary planar scene .
the group velocity @xmath213 induced by the rigid - body motion , and such that the dynamics of @xmath36 satisfies , is given by ( * ? ? ?
5.3 ) @xmath214 note that the group velocity @xmath69 induced by camera motion depends on the additional variables @xmath38 and @xmath39 that define the scene geometry at time @xmath215 as well as the scale factor @xmath216 .
since these variables are unmeasurable and can not be extracted directly from the measurements , in the sequel , we rewrite : @xmath217 since @xmath6 is stationary by assumption , the vector @xmath206 can be directly obtained from the set of embedded gyroscopes .
the term @xmath218 is related to the translational motion expressed in the current frame @xmath7 .
if we assume that @xmath219 is constant ( _ e.g. _ the situation in which the camera moves with a constant velocity parallel to the scene or converges exponentially toward it ) , and using the fact that @xmath220 , it is straightforward to verify that @xmath221 , \quad \mbox { with } \gamma=\left ( \frac{v \eta^\top } { d } -\frac{\eta^\top v } { 3 d}i \right ) , \label{gamma}\ ] ] where @xmath222= \gamma\omega_\times -\omega_\times \gamma$ ] is the lie bracket . however , if we assume that @xmath223 is constant ( the situation in which the camera follows a circular trajectory over the scene or performs an exponential convergence towards it ) , it follows that @xmath224 in this section we assume that the group velocity @xmath69 in is not available , but the angular velocity @xmath206 is . the goal is to provide an estimate @xmath225 for @xmath226 to drive the group error @xmath227 to the identity matrix @xmath56 and the error term @xmath228 ( resp .
@xmath229 ) to @xmath230 if @xmath218 ( resp .
@xmath231 ) is constant or slowly time varying .
the observer when @xmath219 is constant is chosen as follows ( compare to ): @xmath232 -k_i{\mbox{\rm ad}}_{\hat{h}^{\top } } \delta(\hat h , p ) .\label{:eq : dot_hat_gamma}\end{aligned}\ ] ] and the observer when @xmath223 is constant is defined as follows : @xmath233 with some positive gain @xmath234 and @xmath111 given by .
[ propo:1 ] consider a camera moving with kinematics and viewing a planar scene .
assume that @xmath2 is stationary and that the orientation velocity @xmath235 is measured and bounded .
let @xmath236 denote the calibrated homography and consider the kinematics along with .
assume that @xmath36 is bounded and that @xmath218 ( resp .
@xmath231 ) is such that it obeys ( resp . ) .
consider the nonlinear observer defined by ( [ : eq : dot_hat_h][:eq : dot_hat_gamma ] ) , ( resp .
( [ : eq : dot_hat_h-2][:eq : dot_hat_gamma_1 ] ) ) along with the innovation @xmath237 given by . then , if the set @xmath80 of the measured directions @xmath112 is consistent , the equilibrium @xmath238 ( resp .
@xmath239 ) is locally asymptotically stable .
we will consider only the situation where the estimate of @xmath218 is used .
the same arguments can also be used for the case where the estimate of @xmath231 is considered .
differentiating @xmath186 and using yields @xmath240 define the following candidate lyapunov function : @xmath241 differentiating @xmath242 and using the fact that @xmath243\right ) \right ) = 0 $ ] , it follows : @xmath244 introducing the above expression of @xmath245 and using the fact that @xmath246 , it follows : @xmath247 finally , introducing the expression of @xmath108 , one gets @xmath248 the derivative of the lyapunov function is negative semi - definite , and equal to zero when @xmath193 .
given that @xmath206 is bounded , it is easily verified that @xmath249 is uniformly continuous and barbalat s lemma can be used to prove asymptotic convergence of @xmath250 .
using the same arguments as in the proof of theorem [ theo:1 ] , it is straightforward to verify that @xmath251 . consequently the left - hand side of the lyapunov expression converges to zero and @xmath252 converges to a constant .
computing the time derivative of @xmath109 and using the fact that @xmath108 converges to zero and @xmath109 converges to @xmath56 , it is straightforward to show that @xmath253 using boundedness of @xmath36 , one ensures that @xmath254 .
the proposed homography observers have been experimentally evaluated using an avt stingray 125b camera and an xsens mti - g imu attached together . this sensor system , depicted in fig .
[ fig : sensorrefimage ] , is also the one used in @xcite .
the imu runs at @xmath255 hz and only the gyroscope measurements are used for feeding the observer .
the camera provides 40 frames per second at a resolution of @xmath256 pixels .
it makes use of a theia sy125 m wide - angle lens whose estimated intrinsic parameters are @xmath257 pixels for the focal length and @xmath258 pixels for the principle point .
the reference target for the visual system is the image shown in fig .
[ fig : sensorrefimage ] printed on a @xmath259 @xmath260 sheet of paper and then placed over a planar surface parallel to the ground . for simplicity , the point - features extracted from the images are only the 4 corners of the target whenever they are visible in the field of view of the camera .
this is the minimum number of points necessary for algebraically reconstructing the homography matrix @xcite .
we have chosen this simplistic scenario as a first experiment to be able to demonstrate the performance of our proposed observer algorithms independent of potential problems with feature detection , matching and tracking in the real image data .
the latter is not the focus of this paper .
the next section describes a more realistic experiment where failures of the image processing stage occur .
there are a plethora of methods for feature detection and feature matching reported in the computer vision literature , the quality of which directly impacts on the precision of the estimated homography .
since this issue is not the main preoccupation of the present study , we have opted to use an efficient second - order minimization ( esm ) dense approach @xcite to directly estimate the homography matrix for simulated `` ground - truth '' comparison and also for computing the image coordinates of the 4 corners of the target .
the experimental results reported next correspond to the observer whose gains are chosen as @xmath261 and @xmath262 .
let us emphasize that measurements of linear velocity are not available for the observer in this experiment .
we have performed a hand - held movement of the camera such that the target s corners frequently exit the camera s field of view .
[ fig : nbrobspoints ] shows that the number of extracted point - features drops below @xmath78 at multiple instances and for extended periods of time .
these critical cases allow us to test the robustness of the proposed observer .
[ fig : costfunction ] shows the time evolution of the cost function @xmath263 related to the homography error , cf . .
the `` ground - truth '' point locations @xmath112 are simulated using the esm homography .
[ fig : estimatedhomography ] illustrates the time evolution of the homography estimated by the proposed observer and of the simulated `` ground - truth '' homography obtained by the esm algorithm .
the first video in the supplementary multimedia material presents the image sequence with highlighted red point - features , a green quadrilateral joining the estimates of the 4 corners of the target , a warped current image , together with the time evolution of the cost function @xmath264 .
the video is also available at @xcite .
observe from figs .
[ fig : costfunction ] and [ fig : estimatedhomography ] that during the initial transient ( before @xmath265s ) the cost function @xmath264 converges quickly and then remains close to zero for the following eleven seconds .
the estimated homography also converges quickly and remains close to the simulated `` ground - truth '' esm homography .
the occasional loss of point - features during that period only marginally affects the overall performance of the proposed observer .
after that first period , there are three periods where the precision of the estimated homography is significantly deteriorated due to the loss of extracted point - features : from @xmath266s to @xmath267s , from @xmath268s to @xmath269s , and particularly from @xmath270s to @xmath271s .
whenever the number of extracted point - features is not sufficient ( @xmath272 ) , the estimated homography is updated mainly based on the forward integration of the kinematic equation using the angular velocity measurements . since the linear velocity is not measured in this experiment , strong divergence of the homography estimation error is almost inevitable in this situation .
this is particularly the case for fast translational camera motion .
however , once the set of extracted point - features becomes consistent again , the estimated homography quickly converges to the simulated `` ground - truth '' homography , demonstrating the robustness of the proposed observer algorithm .
in this section , we present an application of our approach to image stabilization in the presence of very fast camera motion , severe occlusion and specular reflections .
the experiment consists of moving a camera with attached imu rapidly in front of a planar scene as shown in fig .
[ fig : e1 ] , top left .
the aim is to track and stabilize a region of interest in the reference image , shown in fig .
[ fig : e1 ] , bottom left .
the complete result is shown in the second video provided as supplementary material and available at @xcite , which demonstrates highly robust performance under very challenging conditions .
the camera used is a prototype synchronized camera - imu combination with an aptina mt9v034 cmos sensor and an analog devices adis16375 mems imu .
the imu runs at @xmath255 hz , providing an angular velocity measurement to the observer .
the camera provides @xmath273 frames per second at a resolution of @xmath274 pixels .
the estimated camera parameters are @xmath275 pixels for the focal length and @xmath276 pixels for the principle point .
point - features are extracted using the detectsurffeatures routine in matlab r2012b with the standard recommended parameter settings and matched using matlab s klt - based @xcite matchfeatures routine with a matching threshold of @xmath277 .
we have purposefully avoided using more sophisticated image processing routines in order to demonstrate the raw performance of our observer .
it is likely that the results could be further improved by the use of higher quality feature matching algorithms .
it is quite unrealistic to track one and the same set of features through a long video sequence , in particular given the low frame rate and comparatively rapid motion in our test sequence as well as the presence of severe occlusion ( particularly strong in frames @xmath278 and frames @xmath279 ) and large specular reflections ( e.g. frames @xmath280 ) .
we have hence opted to match features between the reference image and each subsequent frame separately . to do this
, we first forward integrate the observer equations using only the gyroscope measurements , i.e. setting the observer gains @xmath114 ( @xmath281 ) and @xmath234 to zero .
we then use the resulting homography estimate to transform the current image before we apply feature extraction and matching .
a klt - based matching algorithm is well suited to this approach since it favors translational motion over rotational motion , and most of the rotational motion has already been compensated for by forward integrating the angular velocity .
we remove remaining @xmath22-to-@xmath265 matches ( an artefact of matlab s klt implementation ) and apply a conservative symmetric bandpass median filter ( 60% passband ) for outlier removal .
again , we have purposefully avoided the use of more sophisticated ( and much more computationally expensive ) alternatives such as ransac @xcite .
finally , we use observer gains of @xmath282 ( @xmath281 ) and @xmath283 to rapidly iterate the observer equations @xmath284 times per video frame . the computational effort for this last step
is negligible compared to the previous image processing steps .
note that the procedure described above does not satisfy the assumptions of our theoretical convergence results , not least because different point - features are selected for every video frame .
however , it is intuitively clear that such a procedure will still work as long as the selected feature set provides persistent excitation to the observer . indeed
, our experimental results strongly support this claim . the experimental results ( cf . the second video in the supplementary material and available at @xcite ) show good and robust performance throughout the entire video sequence , including the previously mentioned passages with severe occlusion and large specular reflections .
even when the number of usable point - feature matches drops below four for a number of frames ( e.g. frames @xmath285 ) , or when our algorithm selects a wrong feature match ( e.g. frame @xmath286 ) , the observer continues to track the region of interest well and quickly recovers from any tracking errors .
in this paper we developed a nonlinear observer for a sequence of homographies represented as elements of the special linear group @xmath0 .
the observer directly uses point - feature correspondences from an image sequence without requiring explicit computation of the individual homographies between any two given images and fuses these measurements with measurements of angular velocity from onboard gyroscopes using the correct lie group geometry .
the stability of the observer has been proved for both cases of known full group velocity and known rigid - body velocities only . even if the characterization of the stability domain still remains an open issue ,
simulation and experimental results have been provided as a complement to the theoretical approach to demonstrate a large domain of stability .
a potential application to image stabilization in the presence of very fast camera motion , severe occlusion and specular reflections has been demonstrated with very encouraging results even for a relatively low video frame rate .
this work was supported by the _ french agence nationale de la recherche _ through the anr astrid scar project `` sensory control of aerial robots '' ( anr-12-astr-0033 ) , and the _
chaire dexcellence en robotique rte - upmc _ , and the _
australian research council _ through the arc discovery project dp0987411 `` state observers for control systems with symmetry '' .
special thanks go to s. omari for helping us with the data acquisition for experimental validation .
f. caballero , l. merino , j. ferruz , and a. ollero .
homography based kalman filter for mosaic building .
applications to uav position estimation . in _ proceedings of the ieee international conference on robotics and automation ( icra ) _ , pages 20042009 , 2007 .
h. de plinval , p. morin , p. mouyon , and t. hamel .
visual servoing for underactuated vtol uavs : a linear homography - based approach . in _ proceedings of the ieee international conference on robotics and automation ( icra ) _ , pages 3004 - 3010 , 2011 .
f. dellaert , c. thorpe , and s. thrun .
super - resolved texture tracking of planar surface patches . in _ proceedings of the ieee / rsj international conference on intelligent robots and systems ( iros ) _ , pages 197203 , 1998 .
f. fraundorfer , c. engels , and d. nister .
topological mapping , localization and navigation using image collections . in _ proceedings of the ieee / rsj international conference on intelligent robots and systems ( iros ) _ , pages 38723877 , 2007 .
t. hamel , r. mahony , j. trumpf , p. morin , and m .- d . hua .
homography estimation on the special linear group based on direct point correspondence . in _ proceedings of the 50th ieee conference on decision and control ( cdc ) _ , pages 79027908 , 2011
. b. d. lucas and t. kanade .
an iterative image registration technique with an application to stereo vision . in _ proceedings of the international joint conference on artificial intelligence _ , pages 674679 , 1981 .
e. malis , t. hamel , r. mahony , and p. morin .
dynamic estimation of homography transformations on the special linear group for visual servo control . in _ proceedings of the ieee international conference on robotics and automation ( icra ) _ , pages 14981503 , 2009 .
i. f. mondragn , p. campoy , c. martnez , and m. a. olivares - mndez .
3d pose estimation based on planar object tracking for uavs control . in _ proceedings of the ieee international conference on robotics and automation ( icra ) _ ,
pages 3541 , 2010 .
g. g. scandaroli , p. morin , and g. silveira . a nonlinear observer approach for concurrent estimation of pose , imu bias and camera - to - imu rotation .
in _ proceedings of the ieee / rsj international conference on intelligent robots and systems ( iros ) _ , pages 33353341 , 2011 . | this paper presents a new algorithm for online estimation of a sequence of homographies applicable to image sequences obtained from robotic vehicles equipped with vision sensors .
the approach taken exploits the underlying special linear group structure of the set of homographies along with gyroscope measurements and direct point - feature correspondences between images to develop temporal filter for the homography estimate .
theoretical analysis and experimental results are provided to demonstrate the robustness of the proposed algorithm .
the experimental results show excellent performance even in the case of very fast camera motion ( relative to frame rate ) , severe occlusion , and in the presence of specular reflections . |
given a vector @xmath1 , where @xmath2 is the resource allocated to user @xmath3 , _ how fair _ is it ?
one approach to quantify the degree of fairness associated with @xmath4 is through a fairness measure , which is a function @xmath5 that maps @xmath4 into a real number .
various fairness measures have been proposed throughout the years , e.g. , in @xcite .
these range from simple ones , e.g. , the ratio between the smallest and the largest entries of @xmath4 , to more sophisticated functions , e.g. , jain s index and the entropy function .
some of these fairness measures map @xmath4 to normalized ranges between 0 and 1 , where 0 denotes the minimum fairness , 1 denotes the maximum fairness , often corresponding to an @xmath4 where all @xmath2 are the same , and a larger value indicate more fairness . for example , min - max ratio @xcite is given by the maximum ratio of any two user s resource allocation , while jain s index @xcite computes a normalized square mean .
how are these fairness measure are related ?
is one measures `` better '' than any other ?
what other measures of fairness may be useful ?
an alternative approach that has gained attention in the networking research community since @xcite is the approach of @xmath0-fairness and the associated utility maximization . given a set of feasible allocations ,
a maximizer of the @xmath0-fair utility function satisfies the definition of @xmath0-fairness .
two well - known examples are as follows : a maximizer of the log utility function ( @xmath6 ) is proportionally fair , and a maximizer of the @xmath0-fair utility function with @xmath7 is max - min fair .
more recently , @xmath0-fair utility functions have also been connected to divergence measures @xcite , and in @xcite , the parameter @xmath0 was viewed as a fairness measure in the sense that a fairer allocation is one that is the maximizer of an @xmath0-fair utility function with larger @xmath0 although the exact role of @xmath0 in trading - off fairness and throughput can sometimes be surprising @xcite . while it is often held that @xmath8 is more fair than @xmath6 , which is in turn more fair than @xmath9 , it remains unclear what it means to say that @xmath10 is more fair than @xmath11 . clearly , these two approaches for quantifying fairness are different . on the one hand ,
@xmath0-fair utility functions are continuous and strictly increasing in each entry of @xmath4 , thus its maximization results in pareto optimal resource allocations . on the other hand ,
scale - invariant fairness measures ( ones that map @xmath4 to the same value as a normalized @xmath4 ) are unaffected by the magnitude of @xmath4 , and an allocation that does not use all the resources can be as fair as one that does .
can the two approaches be unified ? to address the above questions , we develop an axiomatic approach to measure fairness .
we discover that a set of five axioms , each of which simple and intuitive , can lead to a useful family of fairness measures .
the axioms are : the axiom of continuity , of homogeneity , of asymptotic saturation , of irrelevance of partition , and of monotonicity . starting with these five axioms , we can _ generate _ a family of fairness measures from a generator function @xmath12 : any increasing and continuous function that leads to a well - defined `` mean '' function , i.e. , any kolmogorov - nagumo function @xcite .
for example , using power functions with exponent @xmath13 as the generator function , we derive a unique family of fairness measures @xmath14 that includes all of the following as special cases , depending on the choice of @xmath13 : jain s index , maximum or minimum ratio , entropy , and @xmath0-fair utility , and reveals new fairness measures corresponding to other ranges of @xmath13 . in particular , for @xmath15 , well - known fairness measures ( e.g. , jain s index and entropy ) are special cases of our construction , and we generalize jain s index to provide a flexible tradeoff between `` resolution '' and `` strictness '' of the fairness measure .
for @xmath16 , @xmath0-fair utility functions can be factorized as the product of two components : our fairness measure with @xmath17 and a function of the total throughput that captures the scale , or efficiency , of @xmath4 .
such a factorization also quantifies a tradeoff between fairness and efficiency in achieving pareto dominance with the maximum possible @xmath0 , and facilitates a clearer understanding of what it means to say that a larger @xmath0 is `` more fair '' for general @xmath18 .
the axiomatic construction of fairness measures also illuminates their engineering implications .
any fairness measure satisfying the five axioms can be proven to have many useful properties , including schur - concavity @xcite .
consequently , any operation balancing resources between two user always results in a higher fairness value , extending previous results using majorization to characterize fairness @xcite .
the development of an axiomatic theory of fairness takes another turn towards the end of the paper . by removing the axiom of homogeneity
, we propose an alternative set of four axioms , which allows efficiency of resource allocation be jointly captured in the fairness measure .
we show how this alternative system connects with constrained optimization based resource allocation , where magnitude matters due to the feasibility constraint and an objective function that favors efficiency .
the rest of this paper is organized as follows : the five axioms for fairness measures are introduced and discussed in section ii .
schur - concavity and other properties are proven for any fairness measure satisfying the five axioms in section iii .
we construct a unique family of fairness measures in section iv and discuss its relation to previous work .
generalized jain s index is revealed from this family of fairness measures in section v. section vi provides a new understanding of @xmath0fairness by establishing a connection of our fairness measure to the @xmath0-fair utility functions . in section
vii we propose a second set of axioms that directly incorporates a notion of efficiency .
concluding remarks are made in section viii . due to
space limitations all proofs can be found in the online full version @xcite , together with a discussion contrasting this paper with the well - known axiomatic theories of nash bargaining solution and shapley value in economics .
main notation is shown in table i. cc + * variable * & * meaning * + + @xmath19 & resource allocation vector of length @xmath20 + @xmath21 & sorted vector with smallest element being first + @xmath22 & sum of all elements of @xmath19 + @xmath23 & fairness measure ( of parameter @xmath13 ) + @xmath24 & generator function + @xmath25 & positive weights for weighted mean + @xmath26 & vector of all ones of length @xmath20 + @xmath27 & vector @xmath19 majorizes vector @xmath28 + @xmath13 & parameter for power function @xmath29 + @xmath30 & @xmath0-fair utility with parameter @xmath0 + @xmath31 & shannon entropy function + @xmath32 & jain s index + @xmath33 & our utility for fairness and efficiency +
let @xmath19 be a resource allocation vector with @xmath20 non - negative elements .
a fairness measure @xmath34 is a mapping from @xmath4 to a real number , i.e. , @xmath35 , for all integer @xmath36 .
we first introduce the following set of axioms about @xmath5 , whose explanations and implications are given next . 1
. _ axiom of continuity .
_ fairness measure @xmath34 is continuous on @xmath37 for all integer @xmath36 .
axiom of homogeneity .
_ fairness measure @xmath34 is a homogeneous function of degree 0 : @xmath38 without loss of generality , for a single user , we take @xmath39 for all @xmath40 , i.e. , fairness is a constant for @xmath41 .
axiom of asymptotic saturation .
_ fairness measure @xmath34 of equal resource allocations eventually becomes independent of the number of users : @xmath42 4 .
_ axiom of irrelevance of partition .
_ if we partition the elements of @xmath19 into two parts @xmath43 $ ] , the fairness index @xmath44 can be computed recursively ( with respect to a generator function @xmath45 ) and is independent of the partition , i.e. , @xmath46 where @xmath47 and @xmath48 denote the sum of resource vectors @xmath49 and @xmath50 respectively , and @xmath45 is a continuous and strictly monotonic function that can generate the following function @xmath51 : @xmath52 with positive weights satisfying @xmath53 such that @xmath51 qualifies as a _
function @xcite of @xmath54 .
axiom of monotonicity .
_ for @xmath55 users , fairness measure @xmath56 is monotonically increasing as the absolute difference between the two elements ( i.e. @xmath57 ) shrinks to zero .
axioms 12 are very intuitive .
the axiom of continuity says that a slight change in resource allocation shows up as a slight change in the fairness measure .
the axiom of homogeneity says that the fairness measure is independent of the unit of measurement or absolute magnitude of the resource allocation . due to the axiom of of homogeneity , for an optimization formulation of resource allocation
, the fairness measure @xmath34 alone can not be used as the objective function if efficiency ( which depends on magnitude @xmath58 ) is to be captured . in section
vi , we will connect this fairness measure with an efficiency measure in @xmath0-fair utility function . in section vii
, we will remove the axiom of of homogeneity and propose an alternative set of axioms , which make measure @xmath34 dependent on both magnitude and distribution of @xmath19 , thus capturing fairness and efficiency at the same time .
axiom 3 is a technical condition used to ensure uniqueness of the fairness measure and invariance under change of variable by fixing a scaling .
for example , suppose @xmath34 is a fairness measure satisfying all axioms ( with respect to a mean function @xmath45 ) except axiom 3 .
it is easy to see that by making a logarithmic change of variables , fairness measure @xmath59 also satisfies all axioms , respect to a mean function @xmath60 , other than axiom 3 .
rcc + initial : & + & @xmath61 & @xmath62 + level 1 : & @xmath63 & @xmath64 + & @xmath61 @xmath62 & @xmath61 @xmath62 + level 2 : & @xmath65 & @xmath66 + so far , none of axioms 13 concerns the _ construction _ of fairness measure as the number of users varies . a hierarchical construction of fairness is used in axiom 4 , which allows us to derive fairness measure @xmath35 of @xmath20 users recursively from lower dimensions , @xmath67 and @xmath68 for integer @xmath69 .
the recursive computation is illustrated by a two - level representation in table [ level ] .
let @xmath70 $ ] and @xmath71 $ ] .
the computation is performed as follows . at level 1 , since the total resource is divided into two chunks , @xmath72 and @xmath72 , fairness across the chunks obtained in this level is measured by @xmath73 . at level 2 ,
the two chunks of resources are further allocated to @xmath74 and @xmath75 users , achieving fairness @xmath76 and @xmath77 , respectively . to compute overall fairness of the resource allocation @xmath78 $ ]
, we combine the fairness obtained in the two levels using a multiplication in equation ( [ a5 ] ) .
as we consider a continuous and strictly increasing generator function @xmath45 , the function ( [ g ] ) is a mean value @xcite for @xmath79 , which represents the average fairness of individual parts of @xmath19 .
the set of generator functions giving rise to the same fairness measures may not unique , e.g. , logarithm and power functions .
the simplest case is when @xmath12 is identity and @xmath80 for all @xmath3 .
a natural choice of the weight @xmath25 in ( [ a5 ] ) is to choose the value proportional to the sum resource of vector @xmath81 .
more generally , we will consider the following weights @xmath82 where @xmath83 is an arbitrary exponent .
when @xmath84 , weights in ( [ w ] ) are equal and lead to an un - weighted mean in axiom 4 . as shown in section 4 , the parameter @xmath85 can be chosen such that the hierarchical computation is independent of partition as stated in axiom 4 . as a special case of axiom 4 ,
if we denote the resource allocation at level 1 by a vector @xmath86 $ ] and if the resource allocation at level 2 are equal @xmath87 , it is straight forward to verify that axiom 4 implies @xmath88 where @xmath89 is the direct product of two vectors . as we will show in section vii , an extension of equation ( [ direct ] ) gives an alternative way of stating axiom 4 and leads to a set of more general axioms on fairness .
axiom 5 is the only axiom that actually involves a _ value _ statement on fairness : when there are just two users , more equalized is more fair .
this axiom specifies an increasing direction of fairness and ensures uniqueness of @xmath34 .
consider the allocation of a unit resource to two users as @xmath90 $ ] .
it is intuitive that fairness strictly improves as @xmath91 , since the difference between the two resource shares tends to be smaller .
this intuition also holds for all existing fairness measures , e.g. , various , spread , deviation , max - min ratio , jain s index , @xmath0-fair utility , and entropy . by definition ,
axioms are true , as long as they are consistent and non - redundant . however , not all sets of axioms are useful : unifying known notions , discovering new measures and properties , and providing important insights .
we start showing the use of the above five axioms with the following existence ( the axioms are consistent ) and uniqueness results
. all proofs can be found at @xcite .
( _ existence .
_ ) there exists a fairness measure @xmath34 satisfying axioms 15 .
furthermore , the fairness achieved by equal - resource allocations @xmath26 is independent of the choice of @xmath45 , i.e. , @xmath92 where @xmath93 is a constant exponent .
( _ uniqueness .
_ ) given a generator function @xmath12 , the resulting @xmath34 satisfying axioms 15 is unique .
we first prove an intuitive corollary from the five axioms that will be useful for the rest of the presentation . _ ( symmetry . )
_ a fairness measure satisfying axioms 15 is symmetric over @xmath19 : @xmath94 where @xmath95 is an arbitrary permutation of indices @xmath96 .
the symmetry property shows that the fairness measure @xmath34 satisfying axioms 15 is irrelevant of labeling of users .
we now make a direct connection of our axiomatic theory to a line of work on measuring statistical dispersion by vector majorization , including the popular gini coefficient @xcite .
majorization @xcite is a partial order over vectors to study whether the elements of vector @xmath19 are less spread out than the elements of vector @xmath28 .
we say that @xmath19 is majorized by @xmath28 , and we write @xmath97 , if @xmath98 ( always satisfied due to axiom 2 ) and @xmath99 where @xmath100 and @xmath101 are the @xmath3th elements of @xmath21 and @xmath102 , sorted in ascending order . according to this definition , among the vectors with the same sum of elements ,
one with the equal elements is the most majorizing vector .
intuitively , @xmath97 can be interpreted as @xmath28 being a fairer allocation than @xmath19 .
it is a classical result @xcite that @xmath19 is majorized by @xmath28 , if and only if , from @xmath19 we can produce @xmath28 by a finite sequence of robin hood operations . and @xmath103 with @xmath104 and @xmath105 , respectively , for some @xmath106 .
in other words , we take from the rich ( @xmath107 ) , and give to the poor ( @xmath108 ) . ]
majorization alone can not be used to define a fairness measure since it is a partial order and fails to compare vectors in certain cases .
still , if resource allocation @xmath19 is majorized by @xmath28 , it is desirable to have a fairness measure @xmath5 such that @xmath109 .
a function satisfying this property is known as schur - concave . in statistics and economics ,
many measures of statistical dispersion are derived as certain extensions of majorization to the space of non - negative vectors , e.g. gini coefficient and robin hood ratio @xcite .
we show that our fairness measure is schur - concave , and therefore can be viewed as a different extension of majorization .
similar to gini coefficient and robin hood ratio , the partial order of majorization is preserved by our fairness measure .
( _ schur - concavity .
_ ) a fairness measure satisfying axioms 15 is schur - concave : @xmath110 next we present several properties of fairness measures satisfying the axioms , whose proofs rely on schur - concavity . _
( equal - resource allocation is fairest . ) _ a fairness measure @xmath34 satisfying axioms 15 is maximized by equal - resource allocations , i.e. , @xmath111[cor : equal ] _ ( collecting a fixed - tax is unfair . ) _ if a fixed amount @xmath112 of the resource is subtracted from each user ( i.e. @xmath113 for all @xmath3 ) , the resulting fairness measure decreases @xmath114 where @xmath112 must be small enough such that all elements of @xmath115 are positive . _
( inactive user achieves no fairness . ) _ when a fairness measure @xmath34 satisfying axioms 15 is generated by by @xmath116 in [ w ] , removing users with zero resources does not change fairness : @xmath117
for any function @xmath45 satisfying the condition in axiom 4 , we can generate from @xmath45 a unique @xmath34 .
such an @xmath34 is a well - defined fairness measure if it also satisfies axioms 15 .
we then refer to the corresponding @xmath45 as a generator of the fairness measure .
function @xmath45 is a generator if there exists a @xmath34 satisfying axioms 15 with respect to @xmath45 .
we note , however , that different generator functions may generate the same fairness measure .
although it is difficult to find the entire set of generators @xmath45 , we have found that many forms of @xmath45 functions ( e.g. , logarithm , polynomial , exponential , and their combinations ) result in fairness measures equivalent to those generated by the family of power functions .
it remains to be determined if all fairness measures satisfying axioms 15 can be generated by power functions .
in this section , we consider power functions , @xmath118 , parameterized by @xmath13 and derive the resulting family of fairness measures , which indeed satisfy all the axioms .
the absolute value ensures that @xmath45 is non - increasing over @xmath119 for @xmath120 , and over @xmath121 for @xmath122 . from here on , we replace equation ( [ a5 ] ) in axiom 4 by @xmath123 where the weights @xmath25 are given by ( [ w ] ) .
( _ fairness measures generated by power functions _ ) for power mean ( @xmath118 with parameter @xmath13 ) , axioms 15 define a unique family of fairness measures as follows @xmath124^{\frac{1}{\beta } } , \ { \rm for } \ \beta r \le 1\end{aligned}\ ] ] @xmath125^{\frac{1}{\beta } } , \ { \rm for } \ \beta r \ge 1,\end{aligned}\ ] ] where @xmath126 is a constant exponent , which determines the growth rate of maximum fairness as population size @xmath20 increases , i.e. @xmath127 for different parameter @xmath13 , the fairness measures derived above are equivalent up to a constant exponent @xmath93 : @xmath128^r({\bf x}),\end{aligned}\ ] ] if we denote @xmath129 as the fairness measure with parameters @xmath13 and @xmath93 . according to theorem 1
, @xmath93 determines the growth rate of maximum fairness as population size @xmath20 increases . without loss of generality , we choose @xmath130 such that the maximum average fairness per user is a constant @xmath131 . from a user s perspective ,
her perception of maximum fairness is independent of the population size of the system . from now on , we will use a unified representation of the constructed fairness measurers : @xmath132^{\frac{1}{\beta } } , \label{r0}\end{aligned}\ ] ] where sign@xmath133 is the sign function .
the special cases are summarized in table [ table : pre ] , where @xmath13 sweeps from @xmath134 to @xmath135 and @xmath31 denotes the entropy function .
for some values of @xmath13 , the corresponding mean function @xmath51 has a standard name , and for some , known approaches to measure fairness are recovered , while for @xmath136 and @xmath137 , new fairness measures are discovered .
for instance , @xmath138 corresponds to a harmonic mean function and results in jain s index .
@xmath139 gives a geometric mean function and shannon entropy .
the fairness measure of an arithmetic mean function is discontinuous at @xmath140 . to illustrate the fairness measures , we consider an example with @xmath141 users sharing @xmath142 units of resource , and plot fairness @xmath143 over @xmath13 for different sample vectors , i.e. , @xmath144 $ ] , @xmath145 $ ] , @xmath146 $ ] , and @xmath147 $ ] , in figure [ fig : sample ] .
the fairness measures are discontinuous at @xmath139 . for different @xmath13
, we also observe that the fairness order is preserved on two sets of sample vectors : @xmath148 this is an immediate result from the schur - concavity proven in theorem 1 .
ccc + * value of @xmath13 * & * our fairness measure * & * known names * + + @xmath149 & @xmath150 & max ratio + @xmath151 & @xmath152^{\frac{1}{\beta}}$ ] & @xmath0-fair utility + @xmath140 & @xmath153 ( discontinuous ) & no name + @xmath154 & @xmath155^{\frac{1}{\beta}}$ ] & @xmath0-fair utility + @xmath156 & @xmath157 & entropy + @xmath136 & @xmath158^{\frac{1}{\beta}}$ ] & no name + @xmath138 & @xmath159 & jain s index + @xmath137 & @xmath158^{\frac{1}{\beta}}$ ] & no name + @xmath160 & @xmath161 & min ratio + the fairness measures @xmath162 in ( [ r0 ] ) corresponding to the generator function @xmath118 satisfies a number of properties , which give interesting engineering implications to our fairness measure . _
( number of inactive users . ) _ the fairness measures in ( [ r0 ] ) also count the number of inactive users in the system .
when @xmath163 , @xmath164 if any user is assigned zero resource .
when @xmath165 , @xmath166 _ ( threshold level of resource . ) _ if we increase resource allocation to user @xmath3 by a small amount @xmath167 , while not changing other users allocation , the fairness measures in ( [ r0 ] ) increases if and only if @xmath168 and @xmath169 .
_ ( lower bound under box - constraints . ) _ if a resource allocation @xmath78 $ ] satisfies box - constraints , i.e. , @xmath170 for all @xmath3 , the fairness measures in ( [ r0 ] ) is lower bounded by a constant that only depends on @xmath171 : @xmath172 where @xmath173 and @xmath174 .
the bound is tight when a @xmath175 fraction of users have @xmath176 and the remaining @xmath177 fraction of users have @xmath178 .
these results provide intuition on how the family of fairness measures may be interpreted and applied . through corollary 5 , by specifying a level of fairness , we can limit the number of starved users in a system .
corollary 6 implies that @xmath179 serves as a threshold for identifying `` poor '' and `` rich '' users , since assigning an additional @xmath167 amount of resource to user @xmath3 improves fairness if @xmath180 , and reduces fairness if @xmath181 .
additionally , this provides intuition into threshold methods for allocating resources serially .
when @xmath138 ( i.e. , harmonic mean is used in axiom 4 ) , we get a scalar multiple of the widely used jain s index @xmath182 . upon inspection of ( [ r0 ] ) and the specific cases noted in table iii , we note that any @xmath183 the range of fairness measure @xmath184 lies between @xmath185 and @xmath20 .
equivalently , we can say that the fairness _ per user _ resides in the interval
@xmath186 $ ] .
when the limit as @xmath156 is considered , the resulting fairness measure can also be shown to have this property .
because @xmath184 for @xmath187 has this characteristic , we refer to this subclass of our family of fairness measures as the generalization of jain s index .
@xmath188 is a generalized jain s index parameterized by @xmath15 .
the common properties of our fairness index proven in section iii and iv carry over to this generalized jain s index . for @xmath138
, @xmath189 reduces to the original jain s index .
( _ monotonicity with respect to @xmath13 .
_ ) the fairness measures in ( [ r0 ] ) is negative and decreasing for @xmath151 , and positive and increasing for @xmath190 : @xmath191 the monotonicity of fairness measures @xmath143 on @xmath192 gives an engineering interpretation of @xmath13 .
figure [ fig : beta ] plots fairness @xmath193 for resource allocation @xmath90 $ ] and different choices of @xmath194 .
the vertical bars in the figure represent the level sets of function @xmath5 , for values @xmath195 . for fixed resource allocations ,
since @xmath5 increases as @xmath13 approaches 1 , the level sets of @xmath5 are pushed toward the region with small @xmath196 ( i.e. , the low - fairness region ) , resulting in a steeper incline in the region . in the extreme case of @xmath140 ,
all level set boundaries align with the y - axis in the plot .
the fairness measure @xmath5 point - wise converges to step functions @xmath197 .
therefore , parameter @xmath13 characterizes the shape of the fairness measures : a smaller value of @xmath198 ( i.e. , @xmath13 closer to 1 ) causes the level sets to be condensed in the low - fairness region . since the fairness measure must still evaluate to a number between 1 and @xmath20 here , the monotonicty and resulting shift in granularity of the fairness measure associated with varying @xmath13 suggests differences in evaluating unfairness . at one extreme ,
@xmath199 any solution where no user receives an allocation of zero is fairest . on the other hand , as @xmath160
the relationship between @xmath184 and @xmath196 becomes linear , suggesting a stricter concept of fairness for the same allocation , as @xmath160 fairness value drops .
therefore , the parameter @xmath13 can tune the generalization of jain s index @xmath5 for different tradeoffs between the resolution and the strictness of fairness measure .
due to axiom 2 , the axiom of homogeneity , our fairness measures only express desirability over the @xmath200-dimension subspace orthogonal to the @xmath201 vector .
hence , they do not capture any notion of efficiency of an allocation .
we focus in this section on the widely applied @xmath0-fair utility function : @xmath202 we first show that the @xmath0-fairness network utility function can be factored into two components : one corresponding to the family of fairness measures we constructed and one corresponding to efficiency .
we then demonstrate that , for a fixed @xmath0 , the factorization can be viewed as a single point on the optimal tradeoff curve between fairness and efficiency .
furthermore , this particular point is one where maximum emphasis is placed on fairness while maintaining pareto optimality of the allocation .
this allows us to quantitatively interpret the belief of `` larger @xmath0 is more fair '' _ across all @xmath203_. re - arranging the terms of the equation in table [ table : pre ] , we have @xmath204 where @xmath205 is the one - dimensional version of the @xmath0fair utility function with @xmath206 . for @xmath207 ,
it is easy to show that our fairness measure @xmath208 , multiplied by a function of throughput @xmath58 , equals @xmath0-fair utility function with @xmath6 .
similarly , for @xmath209 , it equals @xmath0-fair utility function as @xmath8 .
therefore , equation ( [ utility ] ) also holds for proportional fairness at @xmath6 and max - min fairness at @xmath7 .
equation ( [ utility ] ) demonstrates that the @xmath0-fair utility functions can be factorized as the product of two components : a fairness measure , @xmath210 , and an efficiency measure , @xmath205 .
the fairness measure @xmath210 only depends on the normalized distribution , @xmath211 , of resources ( due to axiom 2 ) , while the efficiency measure is a function of the sum resource @xmath58 .
rcc + allocation : & + & @xmath61 & @xmath62 + factorize : & @xmath212 & @xmath58 + & @xmath213 & @xmath213 + measure : & @xmath214 & @xmath205 + & @xmath62 & @xmath61 + combine : & + the factorization of @xmath0-fair utility functions is illustrated in table [ factorization ] and decouples the two components to tackle issues such as fairness - efficiency tradeoff and feasibility of @xmath4 under a given constraint set.for example , it helps to explain the counter - intuitive throughput behavior in @xcite : an allocation vector that maximizes the @xmath0-fair utility with a larger @xmath0 may not be less efficient , because the @xmath0-fair utility incorporates both fairness and efficiency at the same time .
although corollary [ cor : equal ] states equal allocation is fairest , an @xmath0-fair allocation may not have an equal distribution .
this is because the additional efficiency component in ( [ utility ] ) can skew the optimizer ( i.e. , the resource allocation resulting from @xmath0fair utility maximization ) away from an equal distribution . for this to happen there
must exist an allocation that is feasible ( within the constraint set of realizable allocations ) with a large enough gain in efficiency over all equal distribution allocations .
hence , the magnitude of this skewing depends on the fairness parameter ( @xmath206 ) , the constraint set of @xmath4 , and the relative importance of fairness and efficiency .
guided by the product form of ( [ utility ] ) , we consider a scalarization of the maximization of the two objectives : fairness and efficiency : @xmath215 where @xmath216 is fixed , @xmath217 absorbs the exponent @xmath13 in the fairness component of ( [ utility ] ) and is a weight specifying the relative emphasis placed on the fairness , and @xmath218 the use of the log function later recovers the product in the factorization of ( [ utility ] ) from the sum in the scalarized ( [ eq : tradeoff ] ) .
an allocation vector @xmath219 is said to be pareto dominated by @xmath220 if @xmath221 for all @xmath3 and @xmath222 for at least some @xmath3 .
an allocation is called pareto optimal if it is not pareto dominated by any other feasible allocation .
if the relative emphasis on efficiency is sufficiently high , pareto optimality of the solution can be maintained .
to preserve pareto optimality , we require that if @xmath220 pareto dominates @xmath219 , then @xmath223 . _ preserving pareto optimality . _
the necessary and sufficient condition on @xmath224 such that @xmath223 if @xmath220 pareto dominates @xmath219 is @xmath225 [ thm : pareto ] consider the set of maximizers of ( [ eq : tradeoff ] ) for @xmath224 in the range in theorem 6 : @xmath226 when weight @xmath227 , the corresponding points in @xmath228 is most efficient .
when weight @xmath229 , it can be shown that the factorization in ( [ utility ] ) is equivalent to ( [ eq : tradeoff ] ) . therefore , @xmath0-fairness corresponds to the solution of an optimization that places the maximum emphasis on the fairness measure parameterized by @xmath17 while preserving pareto optimality .
allocations in @xmath228 corresponding to other values of @xmath224 achieve a tradeoff between fairness and efficiency , while pareto optimality is preserved .
figure [ fig : example_b ] illustrates an optimal fairness - efficiency tradeoff curve @xmath230 , \
\forall { \bf x}={\rm
arg}\max_{{\bf x}\in \mathbb{r } } \phi_{\lambda}(\mathbf{x } ) , \
\forall \lambda \bigg\}$ ] corresponding to the constraint set shown in figure [ fig : example_a ] .
the set of optimizers @xmath228 in ( [ optimizer ] ) , which is obtained by maximizing pareto optimal utilities ( [ eq : tradeoff ] ) , is shown by curve @xmath231 in figure [ fig : example_b ] . in the previous subsection we demonstrated the factorization ( [ utility ] ) is an extreme point on the tradeoff curve between fairness and efficiency for fixed @xmath17 .
what happens when @xmath0 becomes bigger ?
we denote by @xmath232 the gradient operator with respect to the vector @xmath219 . for a differentiable function ,
we use the standard inner product ( @xmath233 ) between the gradient of the function and a normalized vector to denote the directional derivative of the function . _
( monotonicity of fairness - efficiency reward ratio . )
_ let allocation @xmath219 be given .
define @xmath234 as the vector pointing from the allocation to the nearest fairness maximizing solution .
then the fairness - efficiency reward ratio : @xmath235 is non - decreasing with @xmath0 , i.e. higher @xmath0 gives a greater relative reward for fairer solutions .
[ thm : fairer ] the the choice of direction @xmath236 is a direct result of axiom 2 and corollary 2 , which together imply that @xmath236 is the direction that most increases fairness and is orthogonal to increases in efficiency .
an increase in either fairness or efficiency is a `` desirable '' outcome .
the choice of @xmath0 dictates exactly how desirable one objective is relative to the other ( for a fixed allocation ) .
theorem [ thm : fairer ] states that , with a larger @xmath0 , there is a larger component of the utility function gradient in the direction of fairer solutions , relative to the component in the direction of more efficiency .
notice , however , that comparison must be in terms of the ratio between these two gradient components rather than the magnitude of the gradient , and both fairness and efficiency may increase simultaneously .
this result provides a justification for the belief that larger @xmath0 is `` more fair '' , not just for @xmath237 , but for any @xmath18 .
figure [ fig : ratio ] depicts how this ratio increases with @xmath206 for some examples allocations .
given a set of useful axioms , it is important to ask if other useful axiomatic systems are possible . by removing or modifying some of the five axioms here , for example , axiom 2 that decouples the concern on efficiency from fairness , what kind of fairness measures will result ?
can an alternative set of axioms lead to the construction of fairness measures that do not automatically decouple from the notions of efficiency and feasibility of resource allocation ? in this section , we propose a set of alternative axioms , which includes axioms 15 as a special case .
let @xmath238 be a general fairness measure satisfying four axioms as follows . 1
. _ axiom of continuity .
_ fairness measure @xmath239 is continuous on @xmath37 for all integer @xmath36 .
axiom of asymptotic saturation .
_ fairness measure @xmath34 of equal resource allocations eventually becomes independent of the number of users : @xmath240 3 .
_ axiom of irrelevance of splitting . _ for an allocation vector @xmath241 $ ] , we split each element @xmath107 into multiple elements by a direct product @xmath242 , where @xmath243 is a non - negative vector .
if the splitting vectors have equal weights @xmath244 , the fairness of the new allocation vector @xmath245 $ ] is given by @xmath246 where @xmath53 are positive weights and @xmath45 is a continuous and strictly increasing function .
_ axiom of monotonicity .
_ for @xmath55 users , fairness measure @xmath247 increases as ratio @xmath248 goes to 1 , when sum resource @xmath249 is fixed .
axioms 1@xmath250 and 2@xmath250 remain the same as axioms 1 and 3 before .
axiom 4@xmath250 is equivalent to axiom 5 with the additional qualification that the sum - resource does not change .
this qualification was previously unnecessary due to axiom 2 @xmath251 does not vary with the amount of total resources however , is now required in the new set of axioms .
axiom 3@xmath250 is used to recursively construct fairness measure @xmath239 from lower dimensions and is similar to the axiom 4 .
the vector @xmath252 $ ] can be viewed as a generalized direct product of vector @xmath19 with two different vectors @xmath253 and @xmath254 , which split the resource of each element of @xmath255 to multiple users . if @xmath256 , this splitting reduces to a direct product .
since the axiom of homogeneity is removed , fairness measure @xmath239 depends on the absolute magnitude of resource vector @xmath19 . using axiom 3@xmath250
, we can prove that @xmath239 is a homogeneous function of real degree .
furthermore , the two sets of axioms are equivalent , if the order of homogeneity is zero .
this means that the new axiomatic system is more general than the original one .
( _ existence and uniqueness .
_ ) for each generator @xmath45 , there exists a unique fairness measure @xmath239 satisfying axioms 1@xmath250 4@xmath250 .
we have , @xmath257 where @xmath258 is the degree of homogeneity and @xmath34 is a fairness measure satisfying axioms 15 with respect to the same generator @xmath45 .
while it is easy to verify that some properties , like that of symmetry , in section iii also hold for fairness measure @xmath239 , some properties of fairness measures satisfying axioms 15 are lost in the generalization .
for instance , we can no longer say that equal allocations are best .
when power generators @xmath259 are considered , from axioms 1@xmath250 4@xmath250 we can derive fairness measure @xmath260 , which is parameterized by both @xmath224 and @xmath13 , @xmath261 this unifies our results in sections iv - vi : generalized jain s index is a special case of @xmath260 for @xmath262 and @xmath187 ; fairness measure @xmath263 is a subclass of @xmath260 for @xmath227 ; and @xmath0-utility is obtained for @xmath264 and @xmath265 by comparing ( [ relationship1 ] ) and ( [ eq : tradeoff ] ) .
the degree of homogeneity @xmath266 determines how @xmath260 scales as throughput increases .
the decomposition of fairness and efficiency in section vi is now an immediate consequence from axioms 1@xmath250 4@xmath250 .
there is a useful connection with the characterization of @xmath0-fair utility function in the last section . the absolute value @xmath267 is equivalent to the parameter used for defining the utility function ( [ eq : tradeoff ] ) in section vi.b . from theorem 6
, we can conclude that fairness measure @xmath260 is pareto optimal if and only if @xmath268 for every @xmath13 , there is a minimum degree of homogeneity such that pareto optimality can be achieved .
when inequality ( [ eq : cond ] ) is not satisfied , @xmath260 loses pareto optimality and produces less throughput - efficient solutions if it is used as an objective function in utility optimization .
fairness measures with small degree of homogeneity @xmath266 are more suitable for computing index values of fairness .
the degree of homogeneity of a fairness measure satisfying axioms 1@xmath250 4@xmath250 parameterizes a tradeoff between the concept of fairness and efficiency .
moreover , when power functions are used as generating functions , the degree of homogeneity is equivalent to @xmath269 in ( [ eq : tradeoff ] ) .
therefore , the intuition behind our result on a maximum @xmath267 ( minimum degree of homogeneity ) to ensure pareto optimality can be extended to the general optimization - theoretic approach to fairness , i.e. for a fairness measure @xmath270 generated from any @xmath12 , there is a minimum degree of homogeneity @xmath269 to produce a pareto optimal solution . just like our first set of axioms generalized jain s index and revealed new fairness measures with desirable properties , the second set of axioms offers a rich family of objective functions . to summarize , in removing axiom 2 and adapting the set of axioms accordingly , we have shown that axioms 1@xmath250 4@xmath250 include axioms 15 as special cases .
the resulting measures are now affected by notions of both efficiency and fairness , with the balance between the two governed by the degree of homogeneity @xmath269 .
axiomatic theories often form a stable foundation for wide applications .
famous examples include that of the first order logic and that of nash equilibrium . in network economics ,
two prominent axiomatic theories have been used to study network resource allocation .
the nash bargaining solution @xcite of which proportional fairness is a generalization and which models the bartering of persons with an initial allotment of goods is based on a system of four axioms . in cooperative game theory in the shapley value solution concept @xcite . given the setup of a coalitional game , four axioms uniquely define the shapley value as the solution concept .
in both of these constructions , effiency was a fundamental axiom in defining the solution .
in fact , both these approaches to cooperative allocation hold pareto optimality as an axiom and thus are more akin to the optimization theoretic approach to fairness . the first family of fairness measures , @xmath5 , is confined to homogeneous functions of degree zero , and the resulting measures are an integral part of the @xmath0-fairness utility function .
one might suspect that it is possible to extend our axiomatic structure to the optimization theoretic fairness approach by relaxing the axiom of homogeneity to homogeneous functions of arbitrary degree .
this is indeed the case as developed by the previous section .
an axiomatic approach to the fundamental concept of fairness illuminates many issues in network resource allocation research .
this paper is far from the end of axiomatic _ theories _ of fairness .
one way to re - examine axioms is to refute their corollaries in the context of network resource allocation .
perhaps all @xmath2 being the same should not be a maximizer of fairness measure , and a fairness measure need not be schur - concave . instead ,
making some @xmath2 bigger should be called more fair if the resulting @xmath4 is bigger in all coordinates , i.e. , those contributing to the overall efficiency should `` fairly '' receive more resources .
perhaps the fairness measure should be a function dependent on the feasible region of allocations .
these possibilities mean that alternative sets of axioms of fairness , ones with a value statement different from that in axiom 5 or axiom 4@xmath250 in this paper , deserve further exploration .
we have also assumed that resource is infinitesimally divisible and has no user or time dependency , that the way resource allocation is decided ( e.g. , by a central controller or autonomously ) is irrelevant , and that the actual allocation can be transparently verified .
none of these assumptions is true . removing them
will further enrich axiomatic theories of fairness in resource allocation .
kelly , a. maulloo and d. tan , rate control in communication networks : shadow prices , proportional fairness and stability , " _ journal of the operational research society _ ,
237 - 252 , 1998 .
l. s. shapley , `` a value for n - person games '' , in _ contributions to the theory of games _ , volume ii , by h.w .
kuhn and a.w .
tucker , editors .
annals of mathematical studies v. 28 , pp . 307317 .
princeton university press , 1953 .
t. lan , d. kao , m. chiang , and a. sabharwal an axiomatic approach to fairness in network resource allocation " , _ technical report , available at http://www.princeton.edu/@xmath271tlan/fairness.pdf_ , 2009 .
applying axiom 4 to resource allocation vector @xmath274 with integers @xmath275 , we have @xmath276 where the second step follows from the axiom 2 by letting @xmath277 and the fact that @xmath278 .
equation ( [ ap1 ] ) shows that @xmath279 is an additive number - theoretical function @xcite , i.e. , @xmath280 further , from axiom 3 , we derive @xmath281 = 0 \label{ap3}\end{aligned}\ ] ] using the result in @xcite , equation ( [ ap2 ] ) and ( [ ap3 ] ) implies that @xmath282 must be a logarithmic function .
we have @xmath283 where @xmath93 is a real constant .
this is exactly ( [ growth1 ] ) after taking an exponential on both sides .
now , we use ( [ growth1 ] ) to derive an expression for the fairness measure deductively , starting from @xmath55 users .
let @xmath284 and @xmath285 be two rational numbers , such that @xmath286 and @xmath287 for some positive integers @xmath288 . using axiom 4 and lemma 1 , we have @xmath289 applying axiom 2 to ( [ ap4 ] ) with @xmath290 , we have @xmath291 for a given function @xmath45 , equation ( [ ap5 ] ) defines fairness measure @xmath292 for two users for rational vector @xmath293 $ ] . when vector @xmath293 $ ] is real , by axiom 1 , fairness measure @xmath292 is uniquely determined by a sequence of rational allocation vectors , whose limit is @xmath293 $ ] .
therefore , equation ( [ ap5 ] ) uniquely defines fairness measure @xmath292 for arbitrary real numbers @xmath294 .
suppose that we have an expression for the fairness measure @xmath295 with @xmath296 users .
to derive @xmath297 for @xmath298 users , we use axiom 4 to obtain the following : @xmath299 by induction , equations ( [ ap5 ] ) and ( [ ap6 ] ) together defines fairness measure for all integer @xmath36 , when the mean function @xmath45 is given . if the resulting fairness measure satisfies axioms 15 , it must be unique according to equations ( [ ap5 ] ) and ( [ ap6 ] ) .
this proves the uniqueness in theorem 2 . to prove the existence in theorem 1 ,
we show that there exists a mean function @xmath45 , such that the resulting fairness index in ( [ ap5 ] ) and ( [ ap6 ] ) satisfies axioms 15 .
we choose @xmath300 and proportional weights ( i.e. @xmath301 ) in ( [ w ] ) . from ( [ ap5 ] )
, we derive @xmath302 let @xmath303 be the sum of the first @xmath74 elements in vector @xmath304 $ ] .
then , using ( [ ap6 ] ) inductively , we obtain @xmath305^{\frac{u_k}{u_k+x_{k+1 } } } \nonumber \\ & & \ \ \ \ = \frac{\left(u_k + x_{k+1}\right)^r}{x_{k+1}^{\frac{x_{k+1}}{u_k+x_{k+1}}}\cdot\prod_{i=1}^kx_i^{\frac{rx_i}{u_k+x_{k+1 } } } } \label{ap7}\end{aligned}\ ] ] by rearranging the terms in ( [ ap7 ] ) , we obtain a fairness measure generated by logarithmic function @xmath300 and proportional weights : @xmath306 where @xmath307 is the sum of all elements .
we need to prove that the fairness measure in ( [ ap8 ] ) satisfies axioms 15 .
it is easy to see that axioms 13 are satisfied by the fairness measure in ( [ ap8 ] ) . to verify axiom 4 , we consider partitioning a resource allocation vector @xmath19 of @xmath20 users into two segments : @xmath70 $ ] and @xmath308 $ ] for arbitrary @xmath69 .
let @xmath309 . from ( [ ap8 ] ) , we conclude that @xmath310^{\frac{u_k}{u_n}}\cdot \left[\frac{u_{n - k}^r}{\prod_{i = k+1}^{n}x_i^{\frac{x_i}{u_{n - k}}}}\right]^{\frac{u_{n - k}}{u_n } } \nonumber \\ & & \
= f(u_k , u_{n - k})\cdot e^{\frac{u_k}{u_n}\log f({\bf x}^1)+ \frac{u_{n - k}}{u_n}\log f({\bf x}^2 ) } \nonumber \\ & & \ = f(\sum_{i=1}^kx_i ,
\sum_{i = k+1}^nx_i)\cdot g^{-1}\left ( \sum_{i=1}^2 s_i\cdot g\left(f({\bf x}^{i})\right)\right),\end{aligned}\ ] ] where weights @xmath311 and @xmath312 are proportional to the sum resource in each segment .
this shows that the fairness measure in ( [ ap8 ] ) is irrelevant to partition . to verify axiom 5
, we consider an allocation vector @xmath90 $ ] and compute its fairness measure as follows @xmath313 taking a logarithm on both sides , we have @xmath314 . \label{ap9}\end{aligned}\ ] ] since the right hand side of ( [ ap9 ] ) is the entropy function , we conclude that @xmath315 is monotonically increasing for @xmath316 $ ] and monotonically decreasing for @xmath317 $ ] .
therefore , the fairness measure in ( [ ap8 ] ) satisfies axioms 15 .
assume symmetry holds for @xmath20 users .
let @xmath320 $ ] be a resource allocation vector and @xmath321 be an arbitrary permutation of the indices @xmath322 .
when @xmath323 , applying axiom 4 , we can use equation ( [ eq : symmetry ] ) to show that @xmath324 when @xmath325 , using the same technique , we have @xmath326 then symmetry also holds for @xmath327 users . because vector @xmath19 is majorized by vector @xmath28 , if and only if , from @xmath19 we can produce @xmath28 by a finite sequence of robin hood operations @xcite , where we replace two elements @xmath107 and @xmath103 with @xmath104 and @xmath105 , respectively , for some @xmath106 , it is necessary and sufficient to show that such an robin hood operation always improves a fairness measure defined by axioms 15 . toward this end , we consider partitioning a resource allocation vector @xmath19 of @xmath20 users into two segments : @xmath328 $ ] and @xmath329 $ ] .
let @xmath330 $ ] where @xmath331 $ ] be the vector obtained from @xmath332 by the robin hood operation .
using axiom 4 , we have @xmath333 where the last step follows form
the monotonicity of @xmath12 and the monotonicity of fairness measure with two - users in axiom 5 , i.e. , @xmath334 therefore , if @xmath19 is majorized by @xmath28 , then we have @xmath109 .
the fairness measure is schur - concave .
the proof for corollary 2 is straightforward , because among the vectors with the same sum of elements , one with the equal elements is the most majorizing vector .
let @xmath335 ( which is always satisfied due to axiom 2 ) . the sum of the @xmath336
smallest elements satisfies @xmath337 then , @xmath338 implies @xmath339 , for any resource allocation vector @xmath19 . due to schur - concavity in theorem 3 ,
it is sufficient to prove that collecting fixed - tax leads to a more majorizing allocation vector . from axiom 2 ,
we consider a vector @xmath340 , which achieves the same fairness as @xmath341 , i.e. , @xmath342 where @xmath343 , such that @xmath344 then , for any integer @xmath345 we have @xmath346 where the third step following from the following inequality @xmath347 we have @xmath348 , which implies @xmath349 . without loss of generality , we assume that @xmath272 .
first , we plug into equations ( [ ap5 ] ) and ( [ ap6 ] ) power mean @xmath29 with weights generated by arbitrary @xmath85 .
equation ( [ ap5 ] ) gives the fairness measure for two users : @xmath352 to derive the fairness measure for three users , we consider two different partitions of the resource allocation vector @xmath353 $ ] as @xmath293,[x_3]$ ] and @xmath354,[x_2,x_3]$ ] . using ( [ ap6 ] ) , we obtain two equivalent form of the fairness measure in ( [ th4_1 ] ) and ( [ th4_2 ] ) .
@xmath355^{\frac{1}{\beta } } \nonumber \\ & & = \frac{(x_1+x_2+x_3)^r}{\left((x_1+x_2)^{\rho+\beta r}+x_3^{\rho+\beta r}\right)^{\frac{1}{\beta}}}\cdot\left[\frac{(x_1+x_2)^{\rho+\beta r}(x_1^\rho+x_2^\rho)}{x_1^{\rho+\beta r}+x_2^{\rho+\beta r}}+x_3^{\rho}\right]^{\frac{1}{\beta } } \label{th4_1}\end{aligned}\ ] ] @xmath356^{\frac{1}{\beta } } \nonumber \\ & & = \frac{(x_1+x_2+x_3)^r}{\left(x_1^{\rho+\beta r}+(x_2+x_3)^{\rho+\beta r}\right)^{\frac{1}{\beta}}}\cdot\left[\frac{(x_2+x_3)^{\rho+\beta r}(x_2^\rho+x_3^\rho)}{x_2^{\rho+\beta r}+x_3^{\rho+\beta r}}+x_1^{\rho}\right]^{\frac{1}{\beta } } \label{th4_2}\end{aligned}\ ] ] as in axiom 4 , the fairness measure is irrelevant to partition .
hence , equations ( [ th4_1 ] ) and ( [ th4_2 ] ) should be equivalent for all @xmath357 . comparing the terms in ( [ th4_1 ] ) and ( [ th4_2 ] )
, we must have @xmath358 or @xmath359 .
when @xmath358 , it is easy to see that @xmath360 is constant
. this case is trivial .
we conclude that the fairness measure must have the following form @xmath361 where @xmath126 is a proper exponent .
let @xmath303 be the sum of the first @xmath74 elements in vector @xmath304 $ ] . then , using ( [ ap6 ] ) inductively , we obtain @xmath362^{\frac{1}{\beta } } \nonumber \\ & & \ = { \displaystyle}\frac{\left ( u_k^{1-\beta r}+x_{k+1}^{1-\beta r } \right)^{\frac{1}{\beta}}}{\left(u_k+x_{k+1 } \right)^{\frac{1}{\beta}-r } } \cdot \left[\frac{\sum_{i=1}^{k } x_i^{1-\beta r } + x_{k+1}^{1-\beta r } } { u_k^{1-\beta r}+x_{k+1}^{1-\beta r}}\right]^{\frac{1}{\beta } } \nonumber \\ & & \
= \frac{\left(\sum_{i=1}^{k+1 } x_i^{1-\beta r } \right)^{\frac{1}{\beta}}}{\left(\sum_{i=1}^{k+1 } x_i \right)^{\frac{1}{\beta}-r } } , \label{th4_4}\end{aligned}\ ] ] which is exactly equation ( [ fairness ] ) in theorem 4 .
we still need to prove that the fairness measure in ( [ th4_4 ] ) satisfies axioms 15 .
it is easy to see that axioms 13 are satisfied by the fairness measure in ( [ th4_4 ] ) . to verify axiom 4 , we consider partitioning a resource allocation vector @xmath19 of @xmath20 users into two segments : @xmath70 $ ] and @xmath308 $ ] for arbitrary @xmath69
let @xmath309 from ( [ th4_4 ] ) , we conclude @xmath363^{\frac{1}{\beta } } \nonumber \\ & & \ = f(u_k , u_{n - k})\cdot \left [ \frac { u_k^{\rho } \frac{\sum_{i=1}^{k } x_i^{1-\beta r}}{u_k^{1-\beta r } } + u_{n - k}^{\rho } \frac{\sum_{i = k+1}^{n } x_i^{1-\beta r}}{u_{n - k}^{1-\beta r } } } { u_k^{\rho } + u_{n - k}^{\rho } } \right]^{\frac{1}{\beta } } \nonumber \\ & & \ = f(\sum_{i=1}^kx_i ,
\sum_{i = k+1}^nx_i)\cdot g^{-1}\left ( \sum_{i=1}^2 s_i\cdot g\left(f({\bf x}^{i})\right)\right),\end{aligned}\ ] ] where weights @xmath364 and @xmath365 are proportional to some power of the sum resource in each segment .
this proves that the fairness measure in ( [ th4_4 ] ) is irrelevant to partition . to verify axiom 5
, we consider an allocation vector @xmath90 $ ] and compute its fairness measure as follows @xmath366^{\frac{1}{\beta}}.\end{aligned}\ ] ] it is easy to verify that when @xmath367 , the fairness measure @xmath315 is increasing for @xmath316 $ ] and decreasing for @xmath317 $ ] .
axiom 5 is satisfied given @xmath367 . putting all conditions in the proof together , we conclude that , when @xmath368 , the fairness measure given by ( [ th4_4 ] ) is positive and satisfies axioms 15 .
similarly , when @xmath369 , the fairness measure given by ( [ fairness ] ) is negative .
the proof for this case is the same and not repeated here .
when @xmath370 is negative , it is easy to show that @xmath164 if @xmath371 .
when @xmath372 , suppose that @xmath74 users are inactive .
from equation ( [ growth1 ] ) and corollaries 1 and 3 , we have @xmath373 which gives @xmath374 .
further , since the number of active users @xmath375 is upper bounded by @xmath34 , the maximum resource is lower bounded by @xmath376 .
let @xmath377 be an auxiliary function , such that @xmath378 since @xmath34 is differentiable , we have @xmath379 \nonumber\end{aligned}\ ] ] because @xmath380 is positive , @xmath381 has a single root at @xmath382 it is straightforward to show that for any @xmath383 , we have @xmath384 therefore , when @xmath108 @xmath385 are fixed , @xmath34 is maximized by @xmath386 . to derive an lower bound on @xmath34 under the box constraints @xmath170 @xmath387
, we first argue that @xmath34 is minimized only if users are assigned resource @xmath388 or @xmath389 . using the box constraints and corollary 6 , we have @xmath390 similarly
, we can show @xmath391 according to axiom 4 , @xmath34 is increasing on @xmath392 $ ] and decreasing on @xmath393 $ ] .
hence , @xmath34 is minimized only if all @xmath107 take the boundary values in the box constraints , i.e. , @xmath394 let @xmath173 and @xmath175 be fraction of users who receive @xmath389 . by relaxing the constraint @xmath395 to @xmath396 $ ] , we derive an lower bound on @xmath34 as follows @xmath397,\forall i } f({\bf x } ) \nonumber \\ & & \ \ = \min_{\mu\in\left\{\frac{i}{n},\forall i\right\ } } { \rm sign}(1-\beta)\cdot n\left[\frac { \mu\gamma^{1-\beta}+(1-\mu)}{\left(\mu\gamma+1-\mu\right)^{1-\beta}}\right]^{\frac{1}{\beta } } \nonumber \\ & & \ \
\min_{\mu\in[0,1 ] } { \rm sign}(1-\beta)\cdot n\left[\frac { \mu\gamma^{1-\beta}+(1-\mu)}{\left(\mu\gamma+1-\mu\right)^{1-\beta}}\right]^{\frac{1}{\beta}}. \label{ap12}\end{aligned}\ ] ] to find the minimizer in the last optimization problem above , we first recognize that at the two boundary points @xmath398 and @xmath399 ( i.e. all users receive the same amount of resource ) , @xmath400 achieves its maximum value .
therefore , the minimum value is achieved by some @xmath401 .
if @xmath402 is the minimizer of ( [ ap12 ] ) , it is necessary that the first order derivative of the right hand side of ( [ ap12 ] ) is zero , i.e. , @xmath403}{\partial \mu}=0.\end{aligned}\ ] ] soling the above equation , we obtain @xmath404=\left(\gamma^{1-\beta}-1\right)\left[(\gamma -1)\mu+1\right ] .
\nonumber\end{aligned}\ ] ] because this equation is a linear in @xmath175 , its root @xmath402 is the unique minimizer of ( [ ap12 ] ) : @xmath405 the lower bound in corollary 7 follows by plugging @xmath402 into ( [ ap12 ] ) .
we first prove the monotonicity of @xmath406 for @xmath407 .
consider two different values @xmath408 .
we define the a function @xmath409 for @xmath410 .
since @xmath411 , the function @xmath412 is convex in @xmath413 .
therefore , we have @xmath414^{\frac{1}{\beta_2 } } \nonumber \\ & & = \left[\sum_{i=1}^n \frac{x_i}{\sum_j x_j } \cdot \phi\left(\left(\frac{x_i}{\sum_j x_j}\right)^{-\beta_1}\right)\right]^{\frac{1}{\beta_2 } } \nonumber \\ & & \le \left[\phi\left(\sum_{i=1^n}\frac{x_i}{\sum_j x_j } \left(\frac{x_i}{\sum_j x_j}\right)^{-\beta_1 } \right)\right]^{\frac{1}{\beta_2 } } \nonumber \\ & & = \left[\phi\left(\sum_{i=1}^n \left(\frac{x_i}{\sum_j x_j}\right)^{1-\beta_1 } \right)\right]^{\frac{1}{\beta_2 } } \nonumber \\ & & = \left[\sum_{i=1}^n \left(\frac{x_i}{\sum_j x_j}\right)^{1-\beta_1 } \right]^{\frac{1}{\beta_1 } } \nonumber \\ & & = f_{\beta_2 } ( { \bf x}),\end{aligned}\ ] ] where the third step follows from jensen s inequality and @xmath415 .
this shows that @xmath406 is increasing on @xmath416 . for @xmath417
, we consider @xmath418 .
the function @xmath409 becomes concave .
we have @xmath414^{\frac{1}{\beta_2 } } \nonumber \\ & & = \left[\sum_{i=1}^n \frac{x_i}{\sum_j x_j } \cdot \phi\left(\left(\frac{x_i}{\sum_j x_j}\right)^{-\beta_1}\right)\right]^{\frac{1}{\beta_2 } } \nonumber \\ & & \le \left[\phi\left(\sum_{i=1^n}\frac{x_i}{\sum_j x_j } \left(\frac{x_i}{\sum_j x_j}\right)^{-\beta_1 } \right)\right]^{\frac{1}{\beta_2 } } \nonumber \\ & & = f_{\beta_2 } ( { \bf x}).\end{aligned}\ ] ] where the third step follows from jensen s inequality and @xmath419 .
therefore , @xmath406 is increasing on @xmath420 . for @xmath421
, we consider @xmath422 .
the function @xmath409 is concave .
we have @xmath423^{\frac{1}{\beta_2 } } \nonumber \\ & & = -\left[\sum_{i=1}^n \frac{x_i}{\sum_j x_j } \cdot \phi\left(\left(\frac{x_i}{\sum_j x_j}\right)^{-\beta_1}\right)\right]^{\frac{1}{\beta_2 } } \nonumber \\ & & \ge -\left[\phi\left(\sum_{i=1^n}\frac{x_i}{\sum_j x_j } \left(\frac{x_i}{\sum_j x_j}\right)^{-\beta_1 } \right)\right]^{\frac{1}{\beta_2 } } \nonumber \\ & & = f_{\beta_2 } ( { \bf x}).\end{aligned}\ ] ] where the third step follows from jensen s inequality and @xmath419 .
therefore , @xmath406 is decreasing on @xmath424 .
this completes the proof of theorem 5 .
we first assume @xmath425 ( which implies @xmath426 ) and show that the condition @xmath427 is necessary and sufficient for preserving pareto optimality .
the case where @xmath187 can be shown using a completely analogous proof . to show that the condition @xmath427 is sufficient , we consider an allocation @xmath19 and a vector @xmath428 such that @xmath429 for all @xmath3 and @xmath430 . clearly , @xmath431 pareto dominates @xmath219 for @xmath432 .
we now consider the difference between the function ( [ eq : tradeoff ] ) evaluated for these two allocations .
first assume @xmath425 , which implies @xmath426 , and @xmath433 if @xmath434 is also more fair than @xmath219 , then showing @xmath435 is trivial , and the difference between the objective evaluated at the two allocations is strictly positive . therefore , we consider the case where @xmath434 is less fair . continuing from ( [ rst0 ] ) and applying the definition in ( [ r0 ] ) yields @xmath436^{\frac{1}{\beta}}\right)+ \log\left(1+\delta\right ) \nonumber \\ { } & \ \ \ \ \ + \lambda \log\left(\left[\sum_{i=1}^n \left(\frac{x_i}{\sum_j x_j}\right)^{1-\beta } \right]^{\frac{1}{\beta}}\right ) \nonumber \\
= { } & -\frac{\lambda}{\beta } \log\left(\frac{\sum_{i=1}^n \left(x'_i\right)^{1-\beta } } { \sum_{i=1}^n \left(x_i\right)^{1-\beta } } \right ) + \left(1 - \lambda\frac{\beta-1}{\beta}\right)\log\left(1+\delta\right ) .
\label{rst01}\end{aligned}\ ] ] next , we prove that the condition @xmath427 is necessary .
suppose @xmath425 and @xmath442 .
we show that there exists two vectors @xmath19 and @xmath443 , such that @xmath444 , while @xmath434 pareto dominates @xmath219 .
consider an allocation @xmath19 of length @xmath327 , such that @xmath445 for @xmath446 and @xmath447 .
clearly , @xmath219 is pareto dominated by another vector @xmath448 , where @xmath449 for @xmath446 and @xmath450 , for some positive @xmath432 . from the last step of ( [ rst01 ] ) , we have @xmath451 it is straight forward to verify that @xmath444 , if we set @xmath452>0.\ ] ] from the definition of @xmath0-fair utility , we compute the numerator and denominator : @xmath453 and @xmath454 notice that both values are positive .
the ratio between these then is @xmath455 it is easily shown that the factor out fron is strictly positive .
the only component that varies with @xmath13 is the ratio between two weighted averages of the same vector with different weights : @xmath456 that average in the numerator places more weight ( @xmath457 ) on elements that decrease more rapidly ( or increase more slowly for the case @xmath458 ) with @xmath13 , implies that the overall numerator decreases more rapidly ( or increases more slowly ) than the denominator .
therefore , ( [ eq : mean_ratio ] ) is monotonically non - increasing , and theorem [ thm : fairer ] is true .
the continuity of @xmath34 follows directly from that of @xmath239 in axioms 1@xmath250 .
let @xmath460 be a positive real number and @xmath28 be a vector of arbitrary length . to prove homogeneity , we make use of axioms 3@xmath250 : @xmath461 ) \nonumber \\ & & \ \ \ \ = f\left(z , z\right)\cdot g^{-1}\left ( s_1\cdot g\left(f({\bf y})\right ) + s_2\cdot g\left(f({\bf y})\right)\right ) \nonumber \\ & & \ \ \ \ = f\left(z , z\right)\cdot f({\bf y } ) \nonumber \\ & & \ \ \ \
= f(1,1 ) \cdot g^{-1}\left ( s_1\cdot g\left(f(z)\right ) + s_2\cdot g\left(f(z)\right)\right ) \cdot f({\bf y } ) \nonumber \\ & & \ \ \ \
= f(1,1 ) \cdot f({\bf y } ) \cdot f(z ) \nonumber\end{aligned}\ ] ] and similarly , @xmath462 comparing the above two equations , we have @xmath463 when @xmath28 is a scalar , using the result in @xcite , equation ( [ produ ] ) implies that @xmath464 must be a logarithmic function with an exponent @xmath269 .
we have @xmath465 which is a homogenous function of order @xmath269 .
therefore , its normalization @xmath34 in ( [ normalization ] ) is a homogenous function of order zero and satisfies axiom 2 . using the homogeneity property and axioms 2@xmath250 , we obtain @xmath466 this is exactly axiom 3 . from axiom 4@xmath250 m ,
axiom 5 is straight since @xmath467 therefore , monotonicity holds for @xmath315 for @xmath316 $ ] and @xmath316 $ ] , respectively . to prove axiom 4 , we choose @xmath468 and @xmath469 in axiom 3@xmath250 , which results in @xmath470 where the second last step uses the fact that @xmath5 is a homogenous function of order zero
this proves axiom 4 .
if @xmath239 satisfies axioms 1@xmath250 - 4@xmath250 , we have shown that its normalization @xmath471 is a fairness measure satisfying axioms 1 - 5 .
therefore , @xmath472 is homogenous function of order @xmath269 and is given by @xmath473 existence and unique of @xmath239 is straightforward from that of @xmath34 in theorems 1 and 2 . | we present a set of five axioms for fairness measures in resource allocation .
a family of fairness measures satisfying the axioms is constructed .
well - known notions such as @xmath0fairness , jain s index , and entropy are shown to be special cases .
properties of fairness measures satisfying the axioms are proven , including schur - concavity . among the engineering implications
is a generalized jain s index that tunes the resolution of the fairness measure , a new understanding of @xmath0-fair utility functions , and an interpretation of `` larger @xmath0 is more fair '' .
we also construct an alternative set of four axioms to capture efficiency objectives and feasibility constraints . |
the supernova remnant ( snr ) b [email protected] lies to the north and east of the region lh@xmath1 120-n206 in the large magellanic cloud ( lmc ) .
the snr was first identified in radio by @xcite , who referred to it simply as n206 .
although this is more properly the name of the larger complex , for simplicity we shall use the same notation in the following discussion . h@xmath1 images of n206 obtained from the magellanic cloud emission - line survey @xcite show largely shell - like emission , with complex filamentary structure along the limb
@xcite noted that these features were in sharp contrast to the x - ray emission as observed by _
rosat _ , which shows the x - ray emission brightening toward the center of the remnant .
the striking difference between x - ray and optical morphologies led the authors to categorize n206 as a centrally brightened " snr , and to suggest that it might undergo the same physical processes as
mixed morphology " snrs @xcite , which have centrally brightened x - ray emission and a radio shell .
@xcite studied n206 at 5 and 14.7 ghz , and found a spectral index @xmath1 ( as in @xmath2 ) over the remnant of about @xmath00.33 .
this value is fairly flat for a snr ; it is on the borderline between the ranges for a pulsar wind nebula ( pwn ) , for which spectral indices tend to be flatter than @xmath00.4 , and for a typical shell - type remnant , for which spectral indices tend to be between @xmath00.8 and @xmath00.3 @xcite .
low resolution ( several - arcminute scales ) and interference from the nearby region made features in the snr interior difficult to examine .
of particular interest , therefore , are the observations by @xcite of n206 at 3 and 6 cm wavelengths ( 8.0 and 4.8 ghz ) , using the australia telescope compact array ( atca ) with resolutions of 18 and 11 ( hpbw ) , respectively .
they saw emission over the face of the entire snr , somewhat brightened toward the limb .
the spectral index calculated from their findings and those of other radio observations was @[email protected] , in the normal range for filled - center snrs .
@xcite also discerned a
peculiar linear feature " within the snr : a narrow wedge of radio emission aligned in projection with the remnant s center . a spectral index map of the snr showed this feature to have a similar index to the rest of the remnant , @xmath40.2 . in the absence of high - resolution x - ray data , the authors could not isolate a point source , and therefore could only speculate that the feature was produced by a low - mass star or compact object .
" using the opening angle " of the linear feature to calculate the mach number of the presumed object s travel , and the length of the linear feature , the authors obtained an estimate for the snr age of 23,000 yr .
x - ray observations of n206 were made with both the _ chandra _ and _ xmm - newton _ observatories .
_ chandra _ observations used the advanced ccd imaging spectrometer ( acis ) , primarily the s3 back - illuminated chip .
datasets were sequence number 500327 , observations 3848 ( 33.1 ks ) and 4421 ( 32.5 ks ) .
data were reduced following procedures recommended by the chandra x - ray center ( cxc ) and analyzed using the cxc s ciao program and xspec .
the two datasets were filtered for high - background times and poor event grades , resulting in a total good time " interval of 65.6 ks . for each dataset , the 05 pixel randomizations were removed and the subpixel resolution " technique10% without decreasing the overall count rates .
documentation at http://cxc.harvard.edu/cont-soft/software/subpixel_resolution.1.4.html ] was applied , allowing us to more closely examine the structure on small spatial scales .
the filtered datasets were then merged , and spectral results extracted from the merged file . as spectral analysis required us to choose fairly wide regions for reasonable counting statistics , the subpixel resolution " files were not used for this purpose . a background region was taken from an annulus surrounding the snr , and the spectrum of this background region was scaled and subtracted from the source spectra .
individual spectra for regions of interest selected from radio and x - ray images , and the corresponding primary and auxiliary response files , were extracted with the ciao psextract script and analyzed in xspec .
spectra were rebinned by spectral energy to achieve a signal - to - noise ratio of 4 in each bin .
we received the pipeline - processed _ xmm - newton _ data from the science operations centre ( soc ) .
observations were made simultaneously with multiple _ xmm - newton_instruments ; in this paper we will concentrate on the european photon imaging camera ( epic ) mos and pn detectors .
the epic - mos data were taken over two intervals , in 2001 november ( observation i d 089210901 , 41.4 ks ) and 2001 december ( 0089210101 , 14.5 ks ) .
the latter observation also included an epic - pn exposure ( 12.0 ks ) .
initial reduction and analysis were carried out using the science analysis software ( sas ) package provided by the soc , with subsequent spectral analysis in xspec .
the data were filtered to remove high background times or poor event grades , reducing the total good " dataset time to 24.2 ks for the first observation ( epic - mos only ) and 12.0 ks for the second observation ( epic - mos and epic - pn ) .
images and spectra were then extracted from the filtered event files .
background regions immediately surrounding the snr but free of point sources were used to produce background spectra , which were then scaled and subtracted from the source spectra .
joint spectral fits were performed with the data from the two epic - mos observations .
images of n206 were obtained with the _ hubble space telescope _ using the wide field planetary camera 2 ( wfpc2 ) with the f656n ( h@xmath1 ) , f673n ( [ ] @xmath5 6347 , 6371 ) and f502n ( [ ] @xmath5 5007 ) filters .
the southwestern side of the snr was observed in h@xmath1 ( 3@xmath6800 s ) , [ ] ( 3@xmath6800 s ) and [ ] ( 6@xmath6800 s ) , all at the same pointing and position angle , but the northeastern side of the snr was observed only in h@xmath1 ( 3@xmath6800 s ) . the data were reduced using the iraf program and stsdas package .
multiple exposures were combined to remove cosmic - ray events , and the resulting files were bias - subtracted .
the images were divided by exposure times to produce count - rate maps , which were then multiplied by the photflam parameter ( provided in the image header ) to convert these to flux - density maps .
we used the synphot task to determine widths for each filter , and multiplied the flux - density maps by the filter widths to produce flux maps .
the files for the wfc and pc were then mosaicked together for the final images .
images for [ ] and h@xmath1 were clipped at 3@xmath7 of sky background to reduce noise , and then divided to produce a calibrated [ ] /h@xmath1 ratio map .
high - dispersion spectra of n206 were obtained with the echelle spectrograph on the 4 m telescope at cerro tololo inter - american observatory ( ctio ) from two observing runs , 2000 december 6 and 2004 january 14 .
both runs used the 79 line mm@xmath8 echelle grating in the single - order , long - slit observing configuration , in which a flat mirror replaced a cross disperser and a post - slit h@xmath1 filter ( @xmath9 = 6563 , @xmath10 = 75 ) was inserted to isolate a single order .
the red long focus camera and the 2000@xmath62000 site2k#6 ccd were used to record data .
the 24 @xmath11 m pixel size corresponds to roughly 0.082 along the dispersion and 026 perpendicular to the dispersion .
limited by the optics of the camera , the useful spatial coverage is @xmath123@xmath13 .
the spectral coverage is wide enough to include both the h@xmath1 line and the flanking [ ] @xmath56548 , 6583 lines .
a slit width of 250 @xmath11 m ( 164 ) was used and the resultant fwhm of the instrumental profile was @xmath14 km s@xmath8 .
the spectral dispersion was calibrated by a th - ar lamp exposure taken in the beginning of the night , but the absolute wavelength was calibrated against the geocoronal h@xmath1line present in the nebular observations . the journal of echelle observations is given in table [ tab : echobs ] .
we undertook new observations using the parkes 64-m radio telescope at the australia telescope national facility .
we observed n206 for 12 hours on 2003
september 29 using the central beam of the 20-cm multibeam receiver at 1374 mhz on the parkes telescope with the aim of detecting pulsed emission from a small x - ray source .
following the recent successful surveys for pulsars in radio nebulae at parkes @xcite , we recorded total - power signals from 96 frequency channels which provide a bandwidth of 288mhz for two polarizations every 1ms .
the sensitivity of the system was @xmath15jy which is a factor of @xmath122.5 more sensitive than an earlier , large - scale survey of the large magellanic cloud @xcite .
off - line processing was carried out to search for any dispersed , periodic signal using standard procedures with the seek software .
no signal was detected .
the morphology of the snr in the _ xmm - newton _ and _ chandra _ observations is complex , with particular structures revealing themselves variously under _ xmm - newton _ s high sensitivity and _ chandra _ s high resolution . in both cases ( fig .
[ fig : xmm_acis ] ) diffuse emission can be seen over the entire face of the remnant , all the way out to the rim ; and a significant increase in x - ray emission appears toward the center . in the _ xmm - newton _
epic - pn and epic - mos observations there is an extension of this central emission that corresponds well to the position of the radio linear feature " .
this apparent correlation is confirmed by the _ chandra _ observations , in which x - ray emission is clearly seen along the length of the linear feature , broadening in the same region of the wedge " as in radio ( fig . [
fig : acis_atca ] ) .
an x - ray point source appears in the _ chandra _ observation at 05@xmath1632@xmath1703@xmath18 , @xmath0710051 .
no counterpart to this source is found in the optical or radio data .
a search of the simbad database shows these coordinates to be within the error circle for _ einstein _ source 2e 0532.6@xmath07102 @xcite . in the absence of features indicating an association between this source and the snr ,
we presume it to be a background source ; possibly , given its hard x - ray emission , an active galactic nucleus .
we will not include this source in the subsequent discussion . the _ hubble space telescope
_ wfpc2 h@xmath1 mosaic of n206 in figure [ fig : hsthafull ] shows the circular , limb - brightened structure of this snr .
an intricate array of loop - like filaments extends from the outermost limb well within the remnant . toward the southwest of the snr
the h@xmath1 emission becomes more prominent ; this is unsurprising , as in this region the n206 snr may overlap slightly with the larger n206 region .
the [ ] emission , as seen in figure [ fig : hst3band]c , tends to the outer edges of the filaments .
[ ] is a useful tracer " of shock fronts , due to the relatively narrow range of temperatures and densities in which this emission dominates .
thus the tendency of the [ ] to lie in the limbward direction of the filaments implies a fairly regular expansion .
there do not appear to be any counterparts in h@xmath1 or [ ] to the linear feature seen in radio by @xcite .
the h@xmath1 and [ ] morphologies , shown in figure [ fig : hst3band]a and b , are quite similar , including the filamentary structure , as is typical for cooled ( @xmath1210@xmath19 k ) post - shock gas .
spectral fits to the x - ray data from the various instruments ( _ chandra _ acis , _ xmm - newton _ epic - mos and epic - pn ) show that the emission from the snr is dominated by thermal emission ; a number of thermal emission lines are visible ( fig .
[ fig : xray_spec ] ) .
we therefore utilized thermal plasma models to determine the plasma parameters .
as it is uncertain whether the plasma has reached ionization equilibrium , we fit both a non - equilibrium ionization model ( nei ; pshock " in xspec ) and a collisional - ionization equilibrium model ( cie ; mekal " in xspec ) ] .
we selected the following regions for further analysis : ( region 1 ) one comprising the entire snr for comparison with other x - ray studies ; ( 2 ) a region enclosing much of the snr limb , which is expected to be dominated by emission from recently shocked gas ; ( 3 ) a region enclosing the central x - ray brightening but excluding emission thought to be associated with the linear feature " ; ( 4 ) a region covering an apparent wedge - shaped x - ray brightening north and south of the linear feature " , excluding a possible point source ; ( 5 ) a region corresponding to the linear feature " seen in radio ; ( 6 - 8 ) three equal - area regions along this feature to examine possible changes over its length ; ( 9 ) a small region surrounding a possible point source at one end of the linear feature " ; ( 10 ) a region covering emission immediately behind the possible point source , excluding a narrow area corresponding to the brightest radio emission from the linear feature " ; and ( 11 ) a region covering the aforementioned radio - bright linear feature .
a list of the regions used for x - ray analysis is given in table [ tab : reglist ] and shown in figure [ fig : xray_regions ] .
we chose two of these regions thought to include only thermal x - ray emission : outer limb " ( region 2 , fig . [ fig : xray_regions]b ) and central " ( region 3 , fig .
[ fig : xray_regions]c ) .
we compared the fits of simple thermal models to the data from each of the x - ray instruments used ( table [ tab : snrspecfit ] ) .
best fits are obtained with sub - solar metal abundances , consistent with the ambient metal abundances in the lmc of about 30% solar @xcite .
fits to the absorption column density are reasonably consistent for regions with good statistics , giving @xmath20 = 3@xmath31 @xmath6 10@xmath21 @xmath22
. fits to the temperature differ by a factor of two depending on whether cie or nei is assumed ; cie fits have @xmath23 = [email protected] kev and nei fits have @xmath23 = [email protected] kev ( table [ tab : snrspecfit ] ) .
the fits to the ionization parameter @xmath24 ( @xmath25 , where @xmath26 is the hot gas density and @xmath27 is the time since ionization ) indicate that the hot gas deviates considerably from ionization equilibrium , with @xmath24 = 3@xmath28 @xmath29 s. in general , the nei model also provides a better statistical fit to the data .
we will therefore refer primarily to the results of the nei fits in the subsequent discussion . in order to facilitate comparison of our x - ray results with other x - ray studies at lower resolution
, we include the whole snr " region ( region 1 in table [ tab : reglist ] , fig .
[ fig : xray_regions]a ) , which includes all x - ray emission from the remnant , including that from the area of the linear feature " .
these fits ( table [ tab : wholesnrspec ] ) largely fall within the range described above for the outer limb " and central " regions , indicating that the snr s emission is dominated by other sources of emission than that of the linear feature . " while an additional power - law component with a normalization of one - third that of the thermal plasma component slightly improves the fit , this improvement is not significant .
the nei fit yields an absorbed flux of 7@xmath30 erg @xmath22 s@xmath8 , an unabsorbed flux of 4@xmath31 erg @xmath22 s@xmath8 , and a luminosity of 8@xmath32 erg s@xmath8 , all over the [email protected] kev range .
the normalization of the nei model fits to x - ray data can be used to calculate the rms electron density ( @xmath33 ) of the hot gas within the snr .
we assume the remnant to be roughly spherical , with a radius of 21 pc , at a distance of 50 kpc .
we further assume that the gas is composed of fully ionized hydrogen and helium ( @xmath34 ) .
then , using the normalization from nei fits to the chandra acis data for the whole snr " region ( table [ tab : wholesnrspec ] ) we find a density in the hot gas of @xmath35 @xmath36 @xmath29 .
using the outer " and central " regions ( table [ tab : outerlimbspec]-[tab : centerspec ] ) , which do not include emission from the area around the linear feature , " for this calculation gives a very similar density of @xmath37 @xmath36 @xmath29 .
the parameter @xmath38 is a volume filling factor for the hot gas .
if we assume @xmath38=1 , to reflect the centrally filled x - ray morphology of the remnant , the hot gas density is @xmath39 @xmath29 ; if the gas only partially fills this volume , the density will rise in inverse proportion to the square root of this filling factor . from this derived density of the x - ray emitting material , we obtain a mass of 5.3@xmath40 @xmath41 g , or 270@xmath310 @xmath41 m@xmath42 .
presuming that the hot gas fills the entire remnant ( @xmath38=1 ) , this mass is greater than that expected from sn ejecta alone , emphasizing that n206 is an older snr whose emission is dominated by swept - up material from the surroundings ; and supports the lmc - like abundances found from the spectral fit .
using the temperature and density from spectral fits to the x - ray data , we find a thermal pressure , @xmath43 , of 3.3@xmath44 @xmath36 dyne @xmath22 in the hot gas . as the analysis in 3.2
will show , this is significantly greater than the thermal pressure we find for the cool shell .
we can also calculate the thermal energy in this gas , @xmath45 .
this gives a thermal energy of 6@xmath46 @xmath41 erg for the hot gas .
finally , we can estimate the shock velocity under the ( somewhat dubious ) assumption that the bulk of the x - ray emission is being generated at the current shock front , using @xmath47 , with the reduced mass @xmath48 ( for he : h = 1:10 ) .
if we presume the newly shocked gas to be at 0.45 kev , we find that such a temperature would result from a shock speed of 620 km s@xmath8 , implying an overall expansion velocity of 470 km s@xmath8 .
for simplicity , we have assumed in the above calculations that the ions have equilibrated with the electrons ; although this assumption may well not be valid , it provides us with at least initial estimates .
the derived energy and pressure are similar to those for other snrs in the adiabatic stage of evolution .
n206 is clearly more complex than the simple shell approximated above .
it features an increase in surface brightness toward the remnant center , a wedge " of emission appearing to surround the linear feature " , and the feature itself , terminating in a possible compact source .
the properties of these features will be discussed in more detail in 4 .
the central brightening on the snr is quite puzzling .
the spectrum of this region ( table [ tab : centerspec ] ) appears similar to that for the limb of the remnant , with slightly higher abundances , particularly oxygen .
there is no evidence for a strong power - law component in this region ; fits with such a component require it to be normalized to a small fraction of the thermal component s emissivity . toward the eastern side of the remnant , surrounding the linear feature , "
the x - rays show a slightly brighter broad wedge " of emission ( region 4 in table [ tab : reglist ] , fig .
[ fig : xray_regions]c ) .
fits to the spectrum of this region ( table [ tab : wedgespec ] ; this spectrum does not include contributions near the compact source " ) suggest a markedly higher plasma temperature than elsewhere in the snr .
while a nonthermal contribution might artificially skew a thermal model fit toward higher temperatures , fits which included nonthermal emission over a tenth of the thermal emissivity could be statistically ruled out at the 90% level .
we also examined a region covering the area of the linear feature " , as seen from radio images ( region 5 in table [ tab : reglist ] , fig .
[ fig : xray_regions]d ) .
we defined a hardness ratio ( h@xmath0s / h+s ) such that s=0.3 - 1.0 kev and h=1.0 - 8.0 kev ( table [ tab : reglist ] ) . using these hardness ratios to compare this region to others throughout the snr
, we find that , while the contribution of soft - x - rays still dominates , the proportion of hard x - rays in this region is greater than elsewhere in the snr .
this increase in the hardness ratio is consistent with the combination of thermal emission from the snr and emission from a harder source , as for example the small - diameter bright source at the tip of this region .
while a localized increase in temperature could also explain this hardening , it would be quite coincidental that this increase appears only in this portion of the snr that also shows the small - diameter x - ray source and elongated radio feature ( fig .
[ fig : xmm_acis]c - e ) .
we performed spectral fits to data from region 5 , though we note that these fits are somewhat limited by the relatively low number of counts .
we find that either a thermal plasma model or a power - law model can provide a statistically adequate fit to this source ( table [ tab : linearspec ] ) .
best " fit arises from a combination of nonequilibrium thermal plasma and power - law models , providing roughly equal contributions to the spectrum .
although the improvement in the fit is not statistically significant , such a combination is consistent with the scenario inferred from the hardness ratios . to examine the detailed properties of the linear feature " , we used the _ chandra _ acis data to analyze emission from smaller segments along that feature ( regions 6 - 8 in table [ tab : reglist ] , fig .
[ fig : xray_regions]e ) .
these fits , summarized in table [ tab : smregspec ] , should be considered as preliminary estimates only .
figure [ fig : hst3band]d shows an [ ] /h@xmath1
ratio map of the southwestern side of n206 .
[ ] /h@xmath1 ratios range from 0.7 to 1.2 across the snr , typical for shocked gas , with an average value of 0.9 .
this is initially somewhat surprising , as it indicates that there has not been much ionization of material from the nearby ob association .
however , maps of the vicinity show an shell , with the n206 region largely situated within the central cavity , so that the shell walls probably block most of the ionizing radiation before it can reach the snr @xcite .
we measured the average h@xmath1 surface brightness of filaments in the flux - calibrated wfpc2 image of n206 to be ( 0.8 - 1.8)@xmath610@xmath49 erg @xmath22 s@xmath8 pix@xmath8 , or ( 0.8 - 1.8)@xmath50 erg s@xmath8 @xmath22 arcsec@xmath51 . assuming a cool shell temperature of 10@xmath19 k
, this surface brightness implies an optical emission measure of 680@xmath3 260 pc @xmath52 .
we presumed the average filament thickness along the line of sight ( 20 @xmath0 30 ) to be equal to its width perpendicular to the line of sight , and used this as a representative number for the path length @xmath53 through the warm ionized gas . using this @xmath53
, we calculate an rms electron density in the shell of about 10@xmath34 @xmath29 .
if we presume the snr to be in the adiabatic point - blast " stage of expansion @xcite , and the cool shell to be representative of the ambient ism , we would expect the ambient ism density to be roughly one - quarter of that in the shell , or about 3@xmath31 @xmath29 . the optical echelle data ( fig . [ fig : echelle ] )
show several lines in the h@xmath1 spectral region , including the narrow geocoronal h@xmath1 ( 6562.85 ) and telluric oh 6 - 1 p2(3.5 ) 6568.779 and 6 - 1 p1(4.5)e / f 6577.183/.386 lines @xcite , as well as broader lines corresponding to doppler - shifted nebular h@xmath1 emission toward the snr .
the latter include both a velocity component constant along the slit , showing the systemic velocity of the snr , and the characteristic bow - shaped pattern deviating from this systemic velocity , showing motions within the expanding gas . in order to measure the systemic velocity , we extracted a velocity profile from a region 13 wide outside of the emission of the snr expansion pattern .
the profile showed two components , one of which was identified as the telluric oh line .
the other component is the h@xmath1 line with a measured doppler shift of 241@xmath34 km s@xmath8 ; as the snr expansion pattern appears to converge to this component , we take this as the systemic velocity for the n206 snr . for comparison
, we note that the nearest position to n206 in the maps of @xcite shows velocity components at 240@xmath37 and 262@xmath326 km s@xmath8 .
to characterize the expansion of the n206 snr , we measure the velocity offsets ( @xmath54 ) from the systemic velocity in both the blue and red directions . as the material at the forefront of the snr expansion
may be reasonably expected to show the highest velocity , we take the greatest of these velocity offsets to represent the overall expansion velocity of the remnant .
the greatest offset in the blue direction ( @xmath55 ) is @xmath0202@xmath35 km s@xmath8 , while the largest @xmath56= @xmath57193@xmath35 km s@xmath8 .
we therefore estimate the expansion velocity @xmath58 = 202@xmath35 km s@xmath8 . using this value for expansion , and assuming that the snr is in the sedov phase
, we would expect the shock velocity @xmath59 = @xmath60 = 270@xmath35 km s@xmath8 .
it should be noted that the errors given here are only the random errors of the measurements .
it is possible that some of the optically emitting material is too faint for detection in these observations .
if the highest - velocity material is undetected , the actual expansion velocity may be higher than our estimate .
for example , @xcite found a higher expansion rate of 250 km s@xmath8 ; the discrepancy may be due to the difficulty of discerning motions in the faint outer material .
to an extent , this may also apply to the large discrepancy between the shock velocity calculated from the temperature of the hot gas in 3.1.1 , and that measured from the warm ionized gas .
more likely , however , this discrepancy is due to two factors .
( 1 ) the x - ray temperature may not be a reliable representation of the newly shocked gas , particularly given its large deviation from a limb - brightened shell morphology .
( 2 ) we expect the x - ray emission to arise from areas where the shock front is moving through diffuse gas , while the optical emission is expected to be generated where the shock is moving through higher density gas , as seen in the relative densities of the hot gas to the warm ionized gas .
thus , we expect that the shock within the optically - emitting clumps has been somewhat slowed by its progress through this denser material .
the overall expansion velocity from the blast wave is probably intermediate between the measured optical expansion of 202 km s@xmath8 and the calculated expansion from x - ray temperatures of 470 km s@xmath8 . using the shell electron density of n=10 @xmath29 calculated above , and presuming @xmath61 and singly ionized he ( @xmath62 )
, we can infer the mass of gas in the shell according to @xmath63 , where @xmath64 is the shell volume . from the filamentary structure at the edge of the shell
, we measure an average shell thickness of about 11@xmath305 ( [email protected] pc at a distance of @xmath1250 kpc to the lmc ) .
if we assume a simple spherical shell of this thickness , we calculate a shell mass of 9@xmath65 g , or 460@xmath3300 m@xmath42 .
this is almost certainly an underestimate , as the filamentary structure shows a much greater extent of cool gas than such a simplistic scenario .
if instead we presume the cool gas occupies a volume filling factor of as much as 0.1 , we obtain a mass estimate of 2.4@xmath66 g , or 1200@xmath3800 m@xmath42 . using these as low- and high - end estimates of the mass range , and the expansion velocity above
, we calculate the kinetic energy in the cool shell , @xmath67 , to find values of 2@xmath68 erg and 5@xmath69 erg , respectively .
one can also use the density calculated above to calculate the thermal pressure in the shell , @xmath70 , where @xmath71 is the boltzmann constant and @xmath72 is the temperature , presumed here to be @xmath73 k. this equation gives a pressure of 3@xmath74 dyne @xmath22 . again , these values for kinetic energy and shell pressure are typical for middle - aged snrs
. the physical characteristics of the n206 snr are summarized in table [ tab : parameters ] .
the high spatial resolution of _ chandra _ s acis allows us to examine specific regions in more detail .
we have isolated three features that appear to be of particular interest : x - ray emission associated with the radio - identified linear feature ; " the x - ray knot and surrounding emission which coincides with the very tip of that linear feature ; and the region of enhanced x - ray surface brightness toward the center of the snr .
below , we describe each of these features individually and then discuss the possible origins for these features .
the interior x - ray emission is brightest from an irregular region extending n@xmath0s for half the diameter of the remnant .
this central emission is clearly dominated by thermal x - rays , with prominent line features including the blends of he - like lines of mg and si .
the emission is largely soft ( @xmath75 2 kev ) and thermal plasma fits to data from this region give parameters for @xmath20 and @xmath23 similar to those found for the snr as a whole .
however , there does appear to be some difference in abundance distributions between the central area and the outer regions of the snr ( excluding emission associated with the linear feature " ) .
the oxygen lines in the central region are more prominent , with respect to the iron blends ( fe l ) , than in other regions of the snr ( table [ tab : centerspec ] ) .
these differences should be treated with caution , as there are large uncertainties in the abundance determinations ; however , the difference in oxygen abundance between the two regions within the snr is significant when compared to the 90% error ranges for this quantity . if the central emission is due to fossil radiation , " i.e. gas that was shock - heated during earlier phases of the snr expansion , we would expect that this gas would have a higher proportion of ejecta to swept - up matter than the more recently - shocked gas , and therefore that the higher oxygen abundance reflects the presence of oxygen - rich ejecta , as seen in the much younger mixed - morphology snr 0103 - 72 in the smc @xcite .
the presence of o - rich material can not be confirmed from _ hubble space telescope _
this probably reflects the fact that the x - ray emission which shows the oxygen excess is primarily from the hot central cavity , where the temperatures are too high and the densities too low for [ ] to show up .
it is worth noting that in addition to the similar oxygen - rich and mixed - morphology snr 0103 - 72 @xcite , the oxygen - rich snr n132d shows , likewise , a higher ratio of oxygen ( o@xmath76 ) to other elements in the remnant interior than at the bright rim @xcite .
nucleosynthesis models indicate that a high oxygen abundance within a snr s ejecta is the result of a type ii sn ( e.g. , * ? ? ?
the examples of mixed - morphology " snrs which show signs of enhanced oxygen abundances , therefore , may imply that the mixed - morphology " snrs are likely to have originated from type ii sne .
as with other mixed - morphology snrs @xcite , the expanding shock of n206 has slowed to the point where bright x - rays from the limb do not dominate the overall emission from the remnant .
thus , it is reasonable to think that n206 follows the typical pattern for mixed - morphology snrs , wherein the central emission is dominated by fossil radiation from earlier large - scale shock heating during the snr s evolution .
an elliptical region ( 65 e@xmath0w @xmath6 21 n@xmath0s ) surrounding the linear feature " in radio was selected from the 6 cm radio image @xcite , and was used to extract spectra from the corresponding x - ray data ( region 5 ) .
comparison of the hardness ratio for this region to those elsewhere in the snr , as discussed in 3.1.2 , suggests the presence of a hard component in addition to the soft emission seen throughout n206 . a combined nei and power - law model with @xmath77 and @xmath78 provided the best fit
. fits to these spectra are summarized in table [ tab : linearspec ] .
( note that the value of @xmath24 obtained for this fit would indicate a plasma in collisional ionization equilibrium ; we retain the nei model , however , for consistency in intercomparison with the other fits . ) using these fitted values , we obtained an absorbed x - ray flux of 4@xmath79 10@xmath80 erg @xmath22 s@xmath8 over the 0.3 - 8.0 kev range , which corresponds to an unabsorbed flux of 1@xmath6 10@xmath81 erg @xmath22 s@xmath8 . at the lmc distance of 50 kpc
, this gives a luminosity of 3@xmath6 10@xmath82 erg s@xmath8 over this energy range .
to further examine the nature of this linear feature , a series of three equal - area regions ( 20@xmath6 15 ) e@xmath0w along the feature were identified in _
chandra _
images for further spectral examination ( regions 6 - 8 in table [ tab : reglist ] ) .
comparison of the hardness ratios defined above for regions 6 - 8 shows a pronounced shift from regions dominated by high - energy photons to those dominated by low - energy photons .
we also performed spectral fits for these regions ( table [ tab : smregspec ] ) .
it should be stressed that , due to the low numbers of counts , these spectral fits are quite uncertain . cited values of @xmath83 are low due to the large error bars for the spectral bins ; however , further binning was deemed undesirable due to the loss of remaining spectral information .
however , as with the hardness ratios discussed above , the fits illustrate the shift from harder to softer emission along the feature .
in order to see whether the relative fluxes of thermal to nonthermal emission would change substantially as one moved from the remnant center toward the x - ray bright knot at the east end of the linear feature , we adopted a dual approach to these spectral fits .
initially , we fit a two - component ( power - law and nei ) model jointly to all three regions , and determined the flux in each component for each region . as an alternate approach
, we fit the power - law component to the high - energy end of the spectrum only ( @xmath842 kev ) and fixed those parameters , then fit the joint model to the low - energy end of the spectrum .
again , the flux in each component was determined .
the results , summarized in table [ tab : linearflux ] , indicate that the x - ray emission along the linear feature is more strongly dominated by nonthermal emission the further one moves eastward ( closer to the knot ) along the feature , and dies off on the west side .
while this result is largely qualitative , given the errors , we argue that the increase in hard " emission toward the knot is most likely to result from a nonthermal source .
the nonthermal spectra of the x - ray emission and radio emission in the linear feature " imply the presence of a pulsar - wind nebula ( pwn ) .
the most likely source for nonthermal x - rays within a snr is synchrotron radiation .
while nonthermal x - ray emission may occasionally be generated by a fast snr shock at the outer limb of a snr , both the relatively slow expansion of this snr and the shape of the emitting region of the nonthermal emission , i.e. over a small spatial segment perpendicular to the shell , argue against this scenario .
likewise , it seems unlikely that the motion of a compact source would be sufficient to produce a bow shock capable of generating significant synchrotron x - rays .
the most plausible explanation , then , is that the acceleration of particles by a
hidden " pulsar is the source of the nonthermal emission ( in both radio and x - rays ) seen in the linear feature .
the westward decrease in the ratio of nonthermal to thermal x - ray flux along the linear feature suggests that it is a pwn generated by an eastward moving pulsar .
the lifetime of the high - energy particles generated in a pwn is comparatively short ; the radio - emitting particles persist and so trail farther behind the x - ray extent .
the radio pulsar observations were unable to find any periodicity greater than 2 ms to the radio signal from the area of the linear feature " .
it is by no means uncommon for a pwn to be detected in radio and x - rays without a radio point - source counterpart ; for example , n157b , also in the lmc , was deduced to have a pwn long before the x - ray pulsar was discovered , and a radio counterpart for that pulsar has yet to be found @xcite .
another possibility that must be considered , of course , is that these features are associated with a background source rather than the snr .
@xcite discuss this scenario , but conclude that this is unlikely .
the presence of extended nonthermal x - ray emission provides an additional reason to believe the feature is not a background source , as such a feature would be highly unusual in a background galaxy .
if we accept the premise that the linear feature " is a pwn , we may anticipate that in part , the morphology of the pwn is created by a bow - shock structure .
the dynamics of a rapidly moving pulsar and evolving snr lead to the prediction of the formation of a bow shock when the snr expansion speed has decelerated sufficiently for the pulsar s motion relative to the shocked snr material to become supersonic ( e.g. * ? ? ?
at this point the pwn is deformed , leading to a substantial offset between the pulsar and the center of the pwn ; this is certainly consistent with the observed bright knot at the outer tip of the linear feature " presumptive pwn .
numerical simulations predict that for a remnant in the sedov stage , this should occur when the distance traveled by the pulsar from the center of the snr , @xmath85 , approaches the snr radius @xmath86 according to @xmath85/@xmath86 @xmath87 0.677 @xcite . presuming that the geometric center of the snr shell accurately represents the site of the sn , the current position of the small - diameter x - ray source is such that the transverse component of @xmath85 is 0.85 @xmath86 , well over the point for bow shock formation .
if the direction of pulsar motion is not perpendicular to the line of sight , @xmath85 increases accordingly .
these findings are in general accord with those of @xcite , and their assumptions in their estimation of the pulsar velocity and age . the fact that emission from the linear feature " in x - ray and radio does not dominate the overall emission from the remnant presents an additional question .
its spectral index would place it among the most flat radio snr shells , although overlapping ( within the error bars ) with remnants such as ic443 @xcite .
it is difficult to see why the snr as a whole should show a radio index typical of filled - center remnants @xcite if the putative pwn is largely confined to this feature .
however , the radio emission from n206 is faint compared to that from other snrs , and therefore the uncertainties in the spectral index determination are considerable .
in addition , residual emission from the n206 region may contaminate the radio emission , adding additional uncertainty . in the _ chandra _ images ,
a distinct knot of x - ray emission appears to the far eastward end of the linear feature , near the snr limb . to investigate whether this knot is consistent with an unresolved source
, we generated a point source of comparable x - ray flux using the marx simulator , and compared the profile of this simulated source to that of the knot .
we conclude that the knot is a small diffuse region of dimension 2 .
this can also by seen by comparison with the appearance of the moderately bright point source @xmath1250 sw of the knot ( identified above with 2e 0532.6 - 7102 ) .
if a point source is embedded within the knot , only @xmath8850% of the counts from the knot could be due to that point source .
we extracted counts within a 2 radius of the knot from the merged _ chandra _ acis events file ( region 9 in table [ tab : reglist ] , fig .
[ fig : xray_regions]e ) , obtaining a count rate after background subtraction of 7.9 @xmath6 10@xmath89 ct s@xmath8 , or 52 counts over the 65.6 ks exposure time , an insufficient quantity to obtain a meaningful spectrum . fig .
[ fig : xmm_acis ] , however , shows that the knot shows noticeably harder emission than that from the surrounding snr . once again using the hardness ratio defined above , ( table [ tab : reglist ] )
, we find a hardness ratio of 0.5 for the knot , in contrast to hardness ratios of -0.45 and -0.58 for the outer " ( region 2 ; snr limb excluding the putative pwn ) and central " ( region 3 ) areas . to obtain a first - order estimate of spectral properties for this knot , we fixed @xmath20 ( and abundances , for the thermal case ) to the values determined by fits to the x - ray emission from the rest of the snr .
the data are consistent with a thermal plasma model with temperatures @xmath84 1 kev , or a power - law model with a spectral index @xmath90 between 2 and 3 . using a power - law model with @xmath91
, we obtain an estimate for the absorbed flux of 6@xmath6 10@xmath92 erg @xmath22 s@xmath8 , an unabsorbed flux of 1@xmath6 10@xmath80 erg @xmath22 s@xmath8 , and a luminosity of 3@xmath6 10@xmath93 erg s@xmath8 at the lmc distance , all over the 0.3 - 8.0 kev energy range .
the maximum luminosity for an embedded point source is therefore @xmath121.5@xmath6 10@xmath93 erg s@xmath8 over this range .
an elliptical region ( 5@xmath68 ) immediately west of the knot , along the linear feature , yielded a count rate of 1.4 @xmath6 10@xmath94 ct s@xmath8 .
this region has a very similar spectrum to that from the knot , suggesting that the spectrum determined from the knot is dominated by emission from its immediate surroundings .
combining the spectra from these two regions allowed us to increase the signal - to - noise ratio for a slightly better spectral fit .
while these fits would still not allow us to rule out a thermal plasma interpretation at 90% confidence , they confirm that a relatively high temperature @xmath95 kev , is required .
more plausible is a power - law fit , with @xmath96 , roughly consistent with that measured from the linear feature " as a whole . while fits to smaller regions along the linear feature suggest a possible spatial variation in the power - law spectral index , the sensitivity of the x - ray observations is insufficient to make this determination at a statistically significant level .
a brief analysis of the x - ray power spectrum shows no evidence for periodic emission from an embedded pulsar .
however , the timing resolution of these observations , 3.2 seconds , would not be sufficient to detect the pulsations from a typical pulsar .
in addition , the estimated maximum of 25 counts from a point source is insufficient to determine a period for its emission , even if the timing resolution were available .
proceeding westward from the knot , x - ray emission associated with the linear feature " broadens into a wedge of brighter x - ray emission which then appears to merge with the central emission .
faint emission connects this patch to the compact knot close to the eastern rim .
we interpret this connection as a bow shock , " clear in the south and barely discernable in the north .
this bow shock merges with the radio linear feature " close to the knot . a bow shock in front of a hypothetical moving pulsar might re - heat the material through which it moves , leaving a trail of fossil radiation " back to the remnant center and merging there with fossil radiation from much earlier fast - moving shocks .
one approach to these features would be to consider much of the x - ray emission between the bright linear feature and the central region as the actual boundary of the bow - shocked gas . in this case
we would have a much larger opening angle for this structure than that found by @xcite . using the x - ray opening angle of about 57@xmath3 5(1.0 @xmath97 rad )
we obtain a mach number @xmath98 = 2.1 @xmath99 .
this is considerably lower than than the @xmath100 value of @xcite , and closer to the theoretical value for a pulsar moving through a snr interior ( @xmath101 , * ? ? ?
we can calculate the sound speed of the hot gas within the snr according to @xmath102 where @xmath103 is the adiabatic index , here taken as 5/3 ; @xmath71 is boltzmann s constant ; @xmath104 is the electron temperature ; @xmath11 is the reduced mass ; and @xmath105 is the hydrogen particle mass . presuming the reduced mass @xmath106 , and using the temperature found above , we obtain a sound speed in the hot gas of 230 @xmath3 110 km s@xmath8 . according to @xmath107 , where @xmath108 is the motion of the pulsar relative to the snr material ,
this gives a mean relative motion of @xmath108=480 @xmath3 240 km s@xmath8 .
if the pulsar is , as it appears , moving radially outward from the snr center , we can presume its motion to be parallel to that of the expanding snr material around the pulsar . in order to obtain a mean relative velocity of @xmath108=480 km s@xmath8 with material moving at the expansion velocity @xmath58 = 202 km s@xmath8 ( slow - expansion case )
, we would require the pulsar to be moving at a speed of roughly 680 km s@xmath8 with respect to the snr center .
if instead we use the x - ray derived expansion speed of @xmath58 = 470 km s@xmath8 ( fast - expansion case ) , the pulsar must be moving with a speed of @xmath12950 km s@xmath8 with respect to the snr center .
if the motion is entirely perpendicular to the line of sight , and the pulsar began at the geometric center of the snr , it would have taken the pulsar about 28,000 yr to reach its current transverse distance of 19.4 pc in the slow - expansion case , and 20,000 yr in the fast - expansion case .
these can , of course , be affected by the viewing angle ; but if we assume the pulsar to be interacting with the snr interior , its maximum distance of travel is the snr radius of 21 pc ( at a 67 angle to the line of sight ) , which puts its time of travel at 30,000 yr in the slow - expansion case and 22,000 yr in the fast - expansion case . in our suggested picture ,
the bow - shocked gas is primarily thermal , representing material within the snr that is shocked to higher temperatures by the encounter with the bow shock .
presumably , the nonthermal x - ray emission is associated with the elongated pwn , as would be the radio emission .
thus we would expect a broad outer cone " of thermal ( shocked snr material ) emission , with an interior cone " of nonthermal ( shocked pwn ) emission .
such a scenario is in accord with hydrodynamic models of bow - shock pwn , in which the mach cone should manifest itself only in the outer bow shock " ( * ? ? ?
* and references therein ) .
the shocked pwn material forms a trail of synchrotron emitting particles opposite the direction of pulsar motion , creating a cometary " or linear " morphology . to examine this picture for consistency with the data , we fit a simple thermal plasma plus power - law model combination to data from two regions within the linear feature : the thin
trail " region , corresponding to the narrow location of the brightest radio feature ( region 11 in table [ tab : reglist ] , fig .
[ fig : xray_regions]f ) , and the bow " region , comprised of emission around the linear feature excluding the aforementioned trail " ( region 10 , fig .
[ fig : xray_regions]f ) . both regions excluded emission within 6 of the presumed pulsar .
the best - fit model combination gave thermal and power - law parameters similar to those given in previous sections ( @xmath109 @xmath22 , @xmath110 kev , @xmath1112.3 ) .
calculating the fraction of the total flux for each model component , we find that the trail " region shows 96% nonthermal flux and only 4% due to thermal plasma , while the bow " region shows 49% thermal flux and 51% nonthermal flux .
errors on these flux ratios , based on the differences in flux due to uncertainties in the fitted parameters at the 90% confidence level , are of order 10%
. alternately , it is possible that some of the x - ray brightening we consider to be associated with the radio linear feature " ( excluding the x - ray knot ) is in fact only a surface brightness fluctuation , and does not accurately represent the mach cone " of the presumptive moving pulsar .
if the mach cone " is more tightly confined than our x - ray based estimate , the resulting mach number may be somewhat higher , up to a value of about @xmath112 when one considers only the angular extent of the bright x - ray emission immediately surrounding the x - ray knot . while it is also possible that the emission from these features is due to the pwn alone , with no substantial bow - shock emission
, the highly elongated morphology and off - center pulsar position argue strongly for the presence of a significant bow - shock contribution .
the ages of supernova remnants are notoriously ill - characterized .
we consider a number of different approaches to this age , in order to find the range of reasonable age estimates . from the fitted x - ray parameters , several estimates are possible .
combining the derived hot gas density with the fitted ionization parameter @xmath24 gives an upper limit age estimate of @xmath1240,000 years for this remnant .
however , an age this large would be difficult to reconcile with the fact that the hot gas in the snr has apparently not yet reached ionization equilibrium . lowering the filling factor shortens this age estimate significantly . a filling factor of 0.25 , for instance , gives an age estimate of 23,000 yr , similar to the estimate based on the radio linear feature " by @xcite .
we can also make another estimate from the sedov relations , in which the relative fractions of thermal and kinetic energies to total energy are constant .
taking the theoretical relation of @xmath113 , our estimate for thermal energy ( for @xmath114 ) would yield an initial explosion energy @xmath115 of @xmath116 erg . assuming the density behind the shock is a factor of 4 greater than that of the unshocked ism
, we obtain an ism mass density @xmath117 of 1.2 @xmath118 g @xmath29 . then from the sedov relation @xmath119 , we have an age of only 9,000 yr .
if the x - ray filling factor is lower than 1 , the age estimate will rise accordingly ; @xmath120 , for instance , gives an age of 17,000 yr .
if we instead presume the overall ambient density to be represented by the current filaments of warm ionized gas , we obtain an ism mass density of 5.3 @xmath121 g @xmath29 .
the sedov relation then gives an age as high as 57,000 yr , demonstrating the critical role played by the ambient density assumption . using the the simple expansion relation @xmath122 , and assuming sedov expansion ( @xmath123=0.4 )
, we can calculate the age of the remnant based on the expansion velocity .
the expansion velocity of 470 km s@xmath8 derived from the x - ray temperature gives a remnant age of 17,000 yr . the expansion velocity obtained from optical echelle spectroscopy of 202 km s@xmath8 , however , leads to a calculated age of 41,000 yr . again
, however , this must be considered an upper limit , as this value of @xmath58 is , as stated above , a lower limit to the blast - wave expansion .
we can try to narrow down these estimates by considering the consistency of the entire picture .
ages above @xmath1240,000 yr are clearly ruled out by the ionization timescale of the x - ray gas and by the observed optical expansion .
the value of 9000 yr from energetics arguments is highly dependent on the assumption of ambient density ; a better lower limit might be that of 17,000 yr from the expansion velocity derived from the x - ray temperature .
when we consider the estimates of pulsar motion ( 4.4 ) we find further limitations : the minimum time for the pulsar to reach its current location is @xmath1220,000 yr , while the maximum is @xmath1230,000 yr . if we choose a velocity intermediate between the x - ray and optically derived values ( e.g. , @xmath12300 km s@xmath8 ) and a hot gas filling factor of 0.25 , we can reconcile most of these numbers .
the ionization timescale for the hot gas then gives an approximate age of 23,000 yr .
pulsar travel times give ages ( depending on viewing angle ) between 24,000 26,000 yr . simple sedov expansion at this velocity gives an age of 27,000 yr .
while arguments from sedov energetics would , for this same filling factor , give us a somewhat lower age of 17,000 yr , this estimate is highly dependent on the assumption used for ambient density , a very uncertain parameter .
we therefore suggest the snr s age most likely falls in the range between 23,000 27,000 yr , with hard " limits of 17,000 yr 40,000 yr .
we have observed the snr in n206 with high - resolution optical and x - ray instruments , and analyzed the results in concert with additional radio data and optical echelle spectra .
we find it highly probable that n206 is the result of a type ii sn , due to its proximity to other massive - star phenomena , enhancement of oxygen abundances , and the presence of a probable compact object .
we use these data to calculate overall properties for the physical components of the snr , and find these typical of a middle - aged snr in the adiabatic stage .
the remnant is over - pressured , and the bulk of the energy budget still resides in thermal energy from the hot interior .
evidently , therefore , the hot gas within the snr still plays a significant role in the remnant s development .
this remnant is particularly unusual for the very different characteristics it displays in different wavelength regimes .
the optical morphology is limb - brightened and highly filamentary ; the radio morphology is center - filled , with diffuse emission over the remnant s face , but some snr shell structure ; and the x - ray morphology appears somewhat centrally brightened .
the best categorization for this remnant appears to be that of mixed - morphology " @xcite , but the picture is complicated by the presence of radio and x - ray emission near the linear feature " .
we analyze the x - ray data for the area surrounding the linear feature " seen in radio .
we find a small , hard x - ray source located at the tip of the radio feature , with a surface brightness profile consistent with the presence of an embedded compact source .
emission from this source and its surroundings is nonthermal , with a power - law index similar to that seen in crab - type objects .
the ratio of nonthermal to thermal x - ray flux decreases with increasing distance from this source .
we conclude that the most probable scenario for this feature is a pulsar moving at moderate velocity through the surrounding snr .
this creates a bow - shock structure in the direction of motion , deforms the surrounding pwn , and leaves behind a trail of synchrotron emission along the line of travel .
however , it would be difficult to attribute most , or even the majority , of the central x - ray emission to that linear feature .
the x - ray emission from the central region is clearly dominated by thermal line emission , rather than nonthermal continuum . the lifetime of relativistic particles energetic enough to generate x - ray emission is not sufficient for those particles to be responsible for the central x - rays , as indicated by the fact that the radio emission , generated by longer - lived particles , shows no particular increase in this central region .
this indicates , somewhat surprisingly , that n206 is both a composite " snr , containing a pwn and shell structure together , and a mixed morphology " snr , with centrally concentrated x - ray emission that can not be simply associated with the pwn .
indeed , n206 is not the only snr to show this combination ; for instance , the galactic remnant w28 combines centrally concentrated x - ray emission with a probable off - center pwn @xcite .
other examples include w44 @xcite and ic443 @xcite .
we shall call this category mixed - composite "
snrs , reflecting the combined action of separate physical processes influencing their emission .
the authors thank the anonymous referee for a very detailed critique of this work , which has much improved it , and also thank brian d. fields for valuable discussions .
rmw and yhc acknowledge support from stsci grant sti go-08110 .
rmw , yhc , and jrd acknowledge support from nasa grant nag 5 - 11159 .
rmw , yhc , fds and jrd acknowledge support from sao grant g03 - 4096 .
mag acknowledges support from the grant aya 2002 - 00376 of the spanish mcyt ( cofunded by feder funds ) .
behar , e. , rasmussen , a. p. , griffiths , r. g. , dennerl , k. , audard , m. , aschenbach , b. , & brinkman , a. c. 2001 , a&al , 365 , 242 bocchino , f. & bykov , a. m. 2001 , a&a , 376 , 248 camilo , f. et al .
2002 , apjl , 567 , 71 camilo , f. et al .
2002 , apjl , 595 , 25 chevalier , r. a. , & liang , e. p. 1989 ,
apj , 344 , 332 chu , y .- h .
, & kennicutt , r. c. 1988 , aj , 95 , 1111 cox , d. , shelton , s. , maciejewski , w. , smith , r. c. , plewa , t. , pawl , a. , & ryczka , m. 1999 , apj , 524 , 179 dunne , b. , chu , y .- h . , & stavely - smith , l. 2005 , in preparation fan , g. l. 2002 , phd thesis , the university of hong kong gaensler , b. , van der swaluw , e. , camilo , f. , kaspi , v. m. , baganoff , f. k. , yusef - zadeh , f. , & manchester , r. n. 2004 , apj , submitted , astro - ph/0312362 harrus , i. m. , hughes , j. p. , & helfand , d. j. 1996 , apjl , 464 , 161 henize , k. g. 1956 , apjs , 2 , 315 kassim , n. e. 1992 , aj , 103 , 943 kawasaki , m. t. et al .
2002 , apj , 572 , 897 klinger , r. j. , dickel , j. r. , fields , b. d. , & milne , d. k. 2002 , aj , 124 , 2135 mathewson , d. s. & clarke , j. n. 1973 , apj , 180 , 725 mcdowell , j. c. 1994 , einstein obs . unscreened ipc data archive , sao head cd - rom series i milne , d. k. , caswell , j. l. , & haynes , r. f. 1980 , mnras , 191 , 469 olbert , c. m. , clearfield , c. r. , williams , n. e. , keohane , j. w. , & frail , d. a. 2001 , apjl , 554 , 205o osterbrock , d. e. , fulbright , j. p. , martel , a. r. , keane , m. j. , trager , s. c. , & basri , g. 1996 , pasp , 108 , 277 park , s. , hughes , j. p. , burrows , d. n. , slane , p. o. , nousek , j. a. , & garmire , g. p. 2003
, apjl , 598 , 95 petre , r. , szymkowiak , a. e. , seward , f. d. , & willingale , r. 1988 , apj , 335 , 215 rho , j. , & borkowski , k. j. 2002 , apj , 575 , 201 rho , j. & petre , r. 1998 , apjl , 503 , 167 rho , j. , petre , r. , schlegel , e. m. , & hester , j. j. 1994 , apj , 430 , 757 rohlfs , k. , kreitschmann , j. , feitzinger , j. v. & siegman , b. c. 1984 , a&a , 137 , 343 russel , s. c. & dopita , m. a. 1992 , apj , 384 , 508 sedov , l. i. 1959 , _ similarity and dimensional methods in mechanics _
( new york : academic ) smith , r. c. & mcels team 1999 , pasa , 15 , 163 trushkin , s. , verkhodanov , o. v. , chernenkov , v. n. , & andernach , h. 2000 , baltic astron .
, 9 , 608 tsujimoto , t. , nomoto , k. , yoshii , y. , hashimoto , m. , yanagida , s. , & thielemann , f .- k .
1995 , mnras , 277 , 945 van der swaluw , e. 2004 , adspr , 33 , 475 van der swaluw , e. , achterberg , a. , gallant , y. a. , downes , t. p. , & keppens , r. 1998 , a&a , 397 , 913 wang , q. d. , gotthelf , e. v. , chu , y .- h . &
dickel , j. r. 2001 , apj , 559 , 275 williams , r. m. , chu , y .- h .
, dickel , j. r. , petre , r. , smith , r. c. , & tavarez , m. 1999 , apjs , 123 , 467 lccccc 2000 dec 6 & h@xmath1 + & 1200 s & e@xmath0w through linear feature " + 2000 dec 6 & h@xmath1 + & 1200 s & n@xmath0s through snr center + 2004 jan 14 & h@xmath1 + & 1200 s & e@xmath0w through bright filament + [ tab : echobs ] lccccc 1 & whole snr & 19,950 & 10,920 & @xmath00.49 + 2 & outer limb & 7,450 & 4,130 & @xmath00.45 + 3 & central & 6,350 & 3,320 & @xmath00.58 + 4 & wedge & 2,375 & 1,400 & @xmath00.46 + 5 & linear feature & 1350 & 430 & @xmath00.24 + 6 & knot & 456 & & 0.26 + 7 & 22 w of knot & 301 & & @xmath00.35 + 8 & 44 w of knot & 264 & & @xmath00.43 + 9 & point src & 94 & & 0.50 + 10 & bow shock & 754 & & @xmath00.45 + 11 & trail & 196 & & 0.076 + [ tab : reglist ] lccccccccccccccccc [ tab : snrspecfit ] pshock & 2.7@xmath124 & 0.47@xmath125 & 0.25@xmath126 & 1.4@xmath127 & 7@xmath128 & 1.38 & 154 + mekal & 3.6@xmath129 & 0.22@xmath130 & 0.14@xmath131 & & 6@xmath132 & 1.93 & 155 + pshock & 7@xmath133 & 0.50@xmath134 & 0.20@xmath135 & 3.4@xmath136 & 8@xmath137 & 1.17 & 83 + mekal & 2.3@xmath138 & 0.24@xmath139 & 0.08@xmath140 & & 0.1@xmath141 & 1.48 & 84 + pshock & 2.2@xmath142 & 0.44@xmath143 & 0.22@xmath144 & 2.1@xmath145 & 7@xmath146 & 1.54 & 241 + & & & & & 1.5@xmath147 + mekal & 3.1@xmath148 & 0.23@xmath149 & 0.12@xmath150 & & 6@xmath137 & 1.96 & 242 + & & & & & 0.12 @xmath3 0.03 + pshock & 3.5 @xmath151 & 0.34 @xmath3 0.08 & 0.25 @xmath3 0.05 & 4 @xmath152 & 1.4 @xmath153 & 3.09 & 91 + mekal & 3.9 @xmath154 & 0.22 @xmath3 0.08 & 0.4 @xmath3 0.2 & & 3 @xmath137 & 3.79 & 92 + pshock & 2.0 @xmath151 & 0.38 @xmath3 0.04 & 0.14 @xmath3 0.03 & 2.7 @xmath155 & 2 @xmath156 & 2.10 & 64 + mekal & 2.7 @xmath154 & 0.22 @xmath3 0.01 & 0.11 @xmath3 0.02 & & 1.0 @xmath157 & 2.98 & 65 + pshock & 2.9@xmath154 & 0.37@xmath3 0.02 & 0.25@xmath3 0.03 & 3.0@xmath158 & 9@xmath159 & 3.21 & 159 + & & & & & 2.1@xmath153 + mekal & 3.6@xmath160 & 0.22@xmath3 0.06 & 0.4@xmath3 0.1 & & 3@xmath161 & 3.97 & 160 + & & & & & 6@xmath137 + [ tab : snrspecfit ] cccccccccccccccccc [ tab : snrspecfit3 ] component & vmekal & vpshock & vpshock & powerlaw + @xmath20 ( @xmath22 ) & 2.8@xmath162 & 2.2@xmath162 & + @xmath23 ( kev ) & 0.264@xmath163 & 0.453@xmath164 & 0.43@xmath165 & + o / o@xmath166 & 0.36@xmath167 & 0.32@xmath168 & 0.30@xmath165 & + ne / ne@xmath166 & 0.24@xmath169 & 0.22@xmath168 & 0.24@xmath165 & + mg / mg@xmath166 & 0.66@xmath170 & 0.39@xmath171 & 0.41@xmath171 & + si / si@xmath166 & 0.9@xmath172 & 0.28@xmath173 & 0.3@xmath174 & + fe / fe@xmath166 & 0.18@xmath165 & 0.20@xmath175 & 0.21@xmath167 & + @xmath24 ( @xmath29 s ) & & 3.5@xmath176 & 3.3@xmath176 & + @xmath90 & & & & 2.2@xmath177 & + acis norm ( @xmath178 ) & 5.25@xmath179 & 1.75@xmath180 & 1.76@xmath180 & 2.3@xmath181 + mos norm ( @xmath178 ) & 1.05@xmath182 & 3.47@xmath183 & 3.59@xmath184 & 7@xmath185 + @xmath83 & 2.73 & 2.23 & + dof & 294 & 294 & + [ tab : wholesnrspec ] cccccccccccccccccc [ tab : snrspecfit3 ] component & vmekal & vpshock & vpshock & powerlaw + @xmath20 ( @xmath22 ) & 2.9@xmath186 & 1.7@xmath162 & + @xmath23 ( kev ) & 0.237@xmath169 & 0.46@xmath168 & 0.46@xmath165 & + o / o@xmath166 & 0.25@xmath187 & 0.22@xmath175 & 0.22@xmath188 & + ne / ne@xmath166 & 0.20@xmath169 & 0.25@xmath189 & 0.25@xmath190 & + mg / mg@xmath166 & 0.7@xmath191 & 0.43@xmath192 & 0.43@xmath193 & + si / si@xmath166 & 0.8@xmath194 & 0.2@xmath195 & 0.2@xmath195 & + fe / fe@xmath166 & 0.21@xmath196 & 0.24@xmath167 & 0.24@xmath167 & + @xmath24 ( @xmath29 s ) & & 2.2@xmath176 & 2.2@xmath176 & + @xmath90 & & & & 2@xmath197 & + acis norm ( @xmath178 ) & 2.69@xmath198 & 5.1@xmath199 & 5.0@xmath200 & 2@xmath201 + mos norm ( @xmath178 ) & 5.8@xmath202 & 1.11@xmath203 & 1.11@xmath204 & 0@xmath205 + @xmath83 & 1.71 & 1.47 & + dof & 203 & 203 & + [ tab : outerlimbspec ] cccccccccccccccccc [ tab : snrspecfit3 ] component & vmekal & vpshock & vpshock & powerlaw + @xmath20 ( @xmath22 ) & 2.8@xmath206 & 3.0@xmath162 & + @xmath23 ( kev ) & 0.278@xmath207 & 0.335@xmath208 & 0.37@xmath168 & + o / o@xmath166 & 0.56@xmath209 & 0.46@xmath210 & 0.46@xmath187 & + ne / ne@xmath166 & 0.16@xmath211 & 0.24@xmath169 & 0.27@xmath187 & + mg / mg@xmath166 & 0.9@xmath191 & 0.5@xmath195 & 0.6@xmath191 & + si / si@xmath166 & 0.7@xmath196 & 0.3@xmath172 & 0.3@xmath172 & + fe / fe@xmath166 & 0.22@xmath188 & 0.21@xmath190 & 0.23@xmath167 & + @xmath24 ( @xmath29 s ) & & 9@xmath212 & 5.5@xmath212 & + @xmath90 & & & & 3@xmath213 + acis norm ( @xmath178 ) & 1.22@xmath214 & 1.13@xmath215 & 7.7@xmath216 & 2@xmath217 + mos norm ( @xmath178 ) & 2.72@xmath218 & 2.5@xmath219 & 1.7@xmath219 & 0@xmath220 + @xmath83 & 2.50 & 2.22 & + dof & 150 & 150 & + [ tab : centerspec ] cccccccccccccccccc [ tab : snrspecfit3 ] component & vmekal & vpshock & vpshock & powerlaw + @xmath20 ( @xmath22 ) & 2.3@xmath221 & 2.2@xmath162 & + @xmath23 ( kev ) & 0.32@xmath165 & 0.90@xmath222 & 0.66@xmath169 & + o / o@xmath166 & 0.31@xmath223 & 0.30@xmath189 & 0.39@xmath171 & + ne / ne@xmath166 & 0.09@xmath190 & 0.12@xmath223 & 0.14@xmath224 & + mg / mg@xmath166 & 0.36@xmath225 & 0.2@xmath174 & 0.4@xmath191 & + si / si@xmath166 & 0.8@xmath196 & 0.2@xmath195 & 0.1@xmath195 & + fe / fe@xmath166 & 0.06@xmath175 & 0.19@xmath190 & 0.19@xmath210 & + @xmath24 ( @xmath29 s ) & & 7@xmath226 & 2.1@xmath227 & + @xmath90 & & & & 1.6@xmath228 & + acis norm ( @xmath178 ) & 5.9@xmath229 & 1.05@xmath230 & 1.14@xmath231 & 2@xmath232 + mos norm ( @xmath178 ) & 4.3@xmath233 & 7.5@xmath234 & 7.9@xmath235 & 2.3@xmath236 + @xmath83 & 1.49 & 1.07 & + dof & 102 & 102 & + [ tab : wedgespec ] cccccccccccccccccc [ tab : snrspecfit3 ] component & vmekal & vpshock & vpshock & powerlaw & powerlaw + @xmath20 ( @xmath22 ) & 1.0@xmath237 & 3.5@xmath238 & & 2.6@xmath239 + @xmath23 ( kev ) & 2.0@xmath240 & 2.8@xmath241 & 0.4@xmath195 & & + o / o@xmath166 & 2@xmath197 & 0.15@xmath242 & 0.2@xmath243 & & + ne / ne@xmath166 & 2@xmath244 & 0.18@xmath245 & 0.2@xmath246 & & + mg / mg@xmath166 & 1@xmath247 & 0.3@xmath195 & 0.7@xmath248 & & + si / si@xmath166 & 0.1@xmath249 & 0.2@xmath250 & 0.4@xmath251 & & + fe / fe@xmath166 & 0.1@xmath252 & 0.11@xmath253 & 0.1@xmath191 & & + @xmath24 ( @xmath29 s ) & & 4@xmath254 & 5@xmath255 & & + @xmath90 & & & & 2.2@xmath256 & 2.8@xmath196 + acis norm & 4.5@xmath257 & 4.2@xmath258 & 3@xmath259 & 1.5@xmath260 & 2.4@xmath261 + mos norm & 1.1@xmath262 & 1.0@xmath200 & 2.0@xmath263 & 2.7@xmath264 & 6@xmath265 + @xmath83 & 1.13 & 1.04 & & 1.08 + dof & 92 & 92 & & 92 + [ tab : linearspec ] lcccccccc knot & 0.9@xmath266 & 4@xmath267 & 1.7@xmath172 & 2.6@xmath268 & 7@xmath269 & 0.89/18 + 22 w of knot & 2.0@xmath270 & 3@xmath226 & 2@xmath271 & 7@xmath269 & 1.2@xmath272 & 0.67/15 + 44 w of knot & 0.36@xmath209 & 8@xmath273 & 3@xmath271 & 2.4@xmath274 & 1.8@xmath275 & 0.78/15 + point source & & & 2.1@xmath276 & & 2.4@xmath277 & 1.4/6 + [ tab : smregspec ] lccccc cool shell & density & 10@xmath34 @xmath29 + cool shell & mass & 160 - 2000 @xmath278 + cool shell & expansion velocity & 202 @xmath35 km s@xmath94 + cool shell & kinetic energy & 1 - 8 @xmath279 erg + cool shell & thermal pressure & 3@xmath280 dyne @xmath22 + hot gas & temperature & [email protected] kev + hot gas & density & 0.24 @xmath281 @xmath29 + hot gas & mass & 270 @xmath310 @xmath282 @xmath278 + hot gas & thermal energy & 6@xmath68 @xmath282 erg + hot gas & thermal pressure & 3.3 @xmath44 @xmath282 dyne @xmath22 + [ tab : parameters ] lccccccccc knot & 2.2@xmath283 & 2@xmath284 & 0.9/0.1 + 22 w of knot & 7@xmath285 & 7.5@xmath286 & 0.5/0.5 + 44 w of knot & 2.7@xmath287 & 7.9@xmath288 & 0.3/0.7 + knot & 2.3@xmath289 & 4.3@xmath290 & 0.8/0.2 + 22 w of knot & 8.3@xmath291 & 7.3@xmath292 & 0.5/0.5 + 44 w of knot & 8@xmath293 & 7.3@xmath291 & 0.1/0.9 + [ tab : linearflux ] | the n206 supernova remnant ( snr ) in the large magellanic cloud ( lmc ) has long been considered a prototypical mixed morphology " snr .
recent observations , however , have added a new twist to this familiar plot : an elongated , radially - oriented radio feature seen in projection against the snr face . utilizing the high resolution and sensitivity available with the _ hubble space telescope _ , _ chandra _ , and _ xmm - newton _ , we have obtained optical emission - line images and spatially resolved x - ray spectral maps for this intriguing snr .
our findings present the snr itself as a remnant in the mid to late stages of its evolution .
x - ray emission associated with the radio linear feature " strongly suggests it to be a pulsar - wind nebula ( pwn ) .
a small x - ray knot is discovered at the outer tip of this feature .
the feature s elongated morphology and the surrounding wedge - shaped x - ray enhancement strongly suggest a bow - shock pwn structure . |
it has long been thought that t hooft s @xmath3 limit @xcite of @xmath4 gauge theory might be usefully described by some sort of string theory .
however , there is an apparently devastating argument , that this `` qcd string '' ( a.k.a . a tower of glueballs ) is _ not _ fundamental string ( of `` string theory '' ) : the graviton appears in the spectrum of the latter , contradicting the well - known folk theorem @xcite forbidding massless spin 2 bound states in a poincar invariant quantum field theory .
a way to evade this argument has been shown by the conjectured equivalence between classical iib superstring theory on an ads@xmath5 background and @xmath6 supersymmetric @xmath7 yang - mills on flat 4 dimensional minkowski space - time @xcite .
the point is that in this example the graviton lives in 5 space - time dimensions and the flat space - time global symmetry ( poincar(3,1 ) ) is only a subgroup of poincar(4,1 ) , which is realized locally , not globally .
thus the `` massless '' 5 dimensional graviton is a composite of the quanta of a flat - space quantum field theory in 4 dimensional space - time .
there is no massless spin 2 particle in this 4d quantum field theory and no folk theorems are violated . in the @xmath6 conformally invariant example , the projection of the 5d graviton onto the 4d minkowski boundary of ads@xmath5 is a multi - gluon continuum state .
but if the mechanism can be extended to the non - conformally invariant gauge theory describing the gluon sector of qcd , the graviton 4d remnant would presumably be a massive spin 2 glueball . to move these statements beyond conjecture ,
one clearly needs to establish the `` dual '' description starting from either of the supposedly equivalent theories .
i think it is clear that the best starting point for such a project is the flat space quantum field theory .
unlike the previous conjectured dualities in string theory , which asserted the equivalence of pairs of poorly defined theories , this duality asserts the equivalence of a poorly defined theory ( string or quantum gravity ) to a perfectly well defined theory ( asymptotically free or conformally invariant quantum gauge theory on flat 4d space - time ) .
indeed , i am inclined to regard this `` duality '' as more analogous to the alternate descriptions of superconductivity given by bcs theory on the one - hand and landau - ginzburg theory on the other .
if this metaphor holds , the flat space quantum field theory should be embraced as the long sought microscopic formulation of string / quantum gravity .
as t hooft showed in his pioneering paper @xcite , the @xmath3 limit of @xmath4 yang - mills theory reduces to a certain sum of planar feynman diagrams .
elegant techniques , involving the explicit elimination of the off - diagonal matrix elements of the matrix field , have been used to obtain this limit in matrix theories of extremely low space - time dimension ( namely d=0,1 ) @xcite , but these methods have failed to deal with theories with space - time dimension @xmath8 . at the moment ,
i see no better approach to the @xmath8 case than setting up a framework to carry out the direct summation of planar graphs . in the mid-1970 s , motivated by the success of light - cone quantization of string theory @xcite , i proposed @xcite that planar diagram sums be carried out by using light - cone parameterization @xmath9 and that a convenient way to digitize the sum was to discretize the momentum conjugate to @xmath10 , @xmath11 with @xmath12 , and imaginary light - cone time @xmath13 , with @xmath14 . in those first papers i restricted attention to scalar field theories , but brower , giles , and
i soon made a first attempt to extend the approach to qcd @xcite . in our setup , the strong t hooft coupling limit @xmath15 favors the fishnet @xcite diagrams that lead to a light - cone string interpretation .
of course , by its very nature a strong coupling limit probes the microscopic details of discretization , and can at best show only rough qualitative resemblance to the continuum theory .
even so , there were a number of loose ends and unsatisfactory features of this first discretization of qcd which needed to be addressed .
motivated by the goal of discovering a more definitive string description of large @xmath0 qcd klaus bering , joel rozowsky , and i set out to remedy these shortcomings , and in this talk i would like to tell you about the results of our efforts @xcite . as you will see
, we have obtained a much improved discretization setup , but have just begun to explore its usefulness in capturing a string picture of the sum of planar diagrams .
an evolving string sweeps out a world sheet @xmath16 in space - time .
one can choose the parameters so that @xmath17 , and so that @xmath18 , the density of @xmath1 , is a constant @xmath19
. then evolution in @xmath2 is generated by the hamiltonian @xmath20,\end{aligned}\ ] ] where for brevity we have called @xmath21 and the prime denotes differentiation with respect to @xmath22 .
of course @xmath23 is the momentum conjugate to @xmath24 .
a key novelty here is that @xmath25 measures the quantity of string present , and its interpretation as a component of momentum is secondary and derivative . in this way
the string is seen to enjoy a galilei invariant dynamics as it moves only in the @xmath26 dimensional transverse space . as shown by mandelstam @xcite ,
interactions are easily introduced by using the path history form of quantum mechanics and including histories in which strings break and join .
technically this is accomplished by first obtaining the imaginary time @xmath27 path integral representation of @xmath28 for free string .
the action in this case is just @xmath29.\end{aligned}\ ] ] here the action is seen to be an integral over a simply connected rectangular domain for an open free string and a cylinder for a closed string .
a diagram describing an arbitrary number of splits and joins is obtained by allowing some number of cuts , each at constant @xmath22 but of varying length , within the domain . the ends of these cuts mark the splitting or joining points , and @xmath30 is discontinuous across them . calling the generic such domain @xmath31 , the complete amplitude is then given by @xmath32
in order to give a nonperturbative definition of the path integrals appearing in the previous section , roscoe giles and i introduced a lattice version @xcite of the domains @xmath31 .
it was only necessary to discretize @xmath33 and @xmath22 .
so we set @xmath34 and @xmath35 , with @xmath36 and @xmath37 fixed positive integers .
then @xmath38 is replaced by @xmath39 , and the functional integration by ordinary integrals . finally , the action is simply replaced by @xmath40 where @xmath41 labels a link on the lattice .
links between nearest neighbor sites in the @xmath33 direction are all present , but those in the @xmath22 direction are occasionally absent reflecting the possibility of splits and joins .
it is a highly nontrivial fact that this apparently noncovariant setup turns out , after the continuum limit , to be fully compatible with poincar invariance in the critical dimension .
to understand how feynman diagrams look in light - cone parameterization , consider the mixed representation of the scalar field propagator : @xmath42 for simplicity we may establish the convention that each line propagates forward in @xmath2 , and correspondingly @xmath43 .
also we may pass to imaginary time , @xmath44 and then write @xmath45 where the second form is in transverse coordinate representation and @xmath46 is the dimensionality of transverse space .
discretizing @xmath47 and @xmath48 , @xmath49 , the coordinate propagator becomes provides a universal ultraviolet cutoff , and every diagram will therefore be finite .
in contrast , the dlcq industry @xcite keeps time continuous , and must regulate ultraviolet divergences in some other fashion . ]
@xmath50 comparing to the previous section , we see that we can crudely think of a planar feynman diagram as a ( coarsely ) discretized world sheet , with a dynamical link dependent string tension @xmath51 .
each link has its own independent @xmath52 which are each summed over all positive integers .
of course only the large fishnet diagrams will bear any actual resemblance to a continuous world sheet !
the sum over all planar diagrams would then define the qcd string dynamics as including an average over all such discretizations ranging from coarse to fine .
also note that a good world sheet path integral should have ( effectively ) only positive weights .
for instance , for @xmath53 theory each vertex contributes a minus sign if @xmath54 .
a good world sheet interpretation requires @xmath55 , the attractive unstable sign .
gauge theories have vertices of both signs , complicating a straightforward world sheet interpretation . to illustrate the effect of our discretization on a standard diagrammatic calculation , consider the sum of those 2 to 2 scattering diagrams in @xmath53 theory shown in fig .
[ twototwo ] .
we work in the transverse center of mass frame .
assume that the ( discrete ) @xmath56 of the initial ( final ) particles is @xmath57 ( @xmath58 ) .
let @xmath59 be the initial energy ( @xmath60 ) .
fix the discrete time of the first vertex at @xmath61 , and sum over all diagrams in which the final particles both propagate to time @xmath62 , multiply by @xmath63 and sum over all @xmath62 from 1 to @xmath64 .
the result for the off - energy shell s - matrix is then @xmath65 compared to standard formal scattering theory , we see that instead of a factor @xmath66 which acts in wavepackets like the standard energy conserving delta function , the use of discrete time has rendered this as @xmath67 .
this replacement is easy to understand because with discrete imaginary time , the amplitudes should be periodic in @xmath68 with period @xmath69 .
it is thus apparent that the scattering amplitude should be identified with the coefficient of this factor . for @xmath55
, the scattering amplitude shows a bound state pole at a real negative value of @xmath68 : @xmath70.\ ] ] in the continuum limit , @xmath71 , one can make the pole location stay finite by tuning @xmath72 to vanish logarithmically as @xmath73 , showing dimensional transmutation in an asymptotically free theory .
the discretization of qcd initially attempted by brower , giles , and me @xcite , was based on a literal transcription of the feynman rules in light - cone gauge .
the transverse gauge field can be described in the complex basis @xmath74 when it takes on the guise of a complex scalar field , described diagrammatically by attaching an arrow to each line of a feynman diagram .
the primitive quartic vertex conserves arrows , but the cubic vertices can act as sources or sinks of arrows .
the longitudinal gauge field @xmath75 does not propagate and can be integrated out to yield an induced quartic vertex , which depends upon the @xmath1 values of the incoming legs in a singular way : @xmath76 upon discretization , we adopted the drastic prescription of simply dropping the infinite contribution at @xmath77 .
furthermore all tadpole diagrams had to be dropped , because of our insistence that no line propagate 0 time steps .
then the strong coupling limit singled out large planar diagrams involving only the primitive quartic couplings , thus leading to an evaluation similar to the @xmath78 example of the previous section .
unfortunately , these quartic couplings have mixed signs : an attractive interaction between gluons of parallel spin and repulsive between gluons of antiparallel spin .
this ferromagnetic interaction pattern meant that our discretization led to a formal strong coupling limit in which the only long string that could form in the limit would have huge total spin .
the essential problem is that attractive interactions between gluons of opposite spin arise in qcd from gluon exchange @xcite , and the discretization chosen in @xcite prevents anti - ferromagnetic gluon exchange from competing at strong coupling with the ferromagnetic quartic interaction .
this , together with our drastic prescription for all of the @xmath79 ills of light - cone quantization , points to the need for a more refined discretized model to adequately describe the strong coupling behavior of large @xmath0 qcd .
clearly what is needed is a prescription that either enhances strong coupling gluon exchange or suppresses the strong coupling quartic interactions .
we found it most natural to arrange the latter by abolishing all quartic interactions , primitive and induced , and replacing them with the exchange of short lived fictitious particles .
this is shown for the primitive quartic interaction in fig.[4vertexto3vertex ] .
the dashed line is associated with the fictitious two - form propagator @xmath80 where the @xmath81 are tunable parameters which are required to vanish rapidly with @xmath82 , the number of time steps propagated .
we treat the induced quartic interaction in a similar fashion , giving the non - dynamical field @xmath75 a short - time propagator @xmath83 the presence of the tunable parameters @xmath84 and @xmath81 is very welcome , because they can be adjusted to arrange the cancelation of cut - off artifacts that can typically spoil poincar invariance in the continuum limit . as a bonus
, we find that our prescription provides the appealing interpretation of tadpole diagrams indicated in fig .
[ tadpole2 ] .
the first two diagrams cancel exactly ( which is fortunate since they ca nt be drawn in our discrete model ! ) leaving the third diagram which _ can _ be drawn .
our complete set of feynman rules is neatly summarized in fig .
[ newrules ] taken from our paper @xcite . [ cols="^,^,^,^ " , ] note that to avoid clutter we have suppressed the double line notation so these rules are completely sufficient for all graphs of planar or cylindrical topology ( @xmath3 ) . for general diagrams , the double line notation must be restored in order to properly account for the @xmath85 corrections .
an easy way to understand why a set of rules with only cubic vertices is possible , is to apply light - cone gauge to the yang - mills lagrangian in first order form .
the upshot is the lagrangian @xmath86\nonumber\\ & & -ig\tr{1\over\partial_-}{\hat a}[a_k,\partial_-a_k ] -ig\tr\phi_{kn}[a_k , a_n],\end{aligned}\ ] ] where @xmath87 .
this makes it clear why we called the fictitious scalar represented by the dashed lines a 2-form : it is a ( pseudo ) scalar only in 3 + 1 dimensions . quite apart from its use as a facilitator for a strong coupling expansion
, our discretization can also serve as a novel way to regulate the diagrams of weak coupling perturbation theory . to illustrate this aspect
, we quote the result for the one loop gluon self - energy , with discretization in place : @xmath88-{(m-1)(11m-1)\over3 m } \right)\nonumber\\ & & -\sum_{k=1}^\infty u^k\sum_{l=1}^{m-1}{lh_k(l)\over mk } -\sum_{k=1}^\infty f_k{u^k\over k } \left(4m[\psi(m)+\gamma]-{7(m-1)\over2}\right ) \bigg ] , \label{pi2new}\end{aligned}\ ] ] where @xmath89 . in order to cancel lattice artifacts in the continuum limit @xmath90 ,
we find the constraints @xmath91 then we obtain @xmath92\ln{q^2\over2mt_0}+{4\over3 } \right\}.\end{aligned}\ ] ] here @xmath93 functions as a uv cut - off . with this understanding
our result for @xmath94 agrees exactly with the known light - cone gauge result @xcite .
notice that the parameters @xmath95 which specify our discretization enter weak coupling physics only through their _ moments _ , for example @xmath96 .
in contrast , the strong coupling limit is sensitive to the values of these parameters at low @xmath62
. thus the two limits give complementary constraints on these parameters .
our main aim in developing this discretization formalism is to establish a framework for handling the sum of all the planar diagrams of @xmath3 gauge theories .
as yet we do nt have any dramatic results to report .
however we have begun studying how the machinery works in simpler situations than full - blown gauge theories . in our paper
@xcite we worked out the sum of the densest ( strong coupling `` fishnet '' ) planar diagrams of @xmath97 scalar field theory .
as expected the result leads to the light - cone quantized bosonic string ( with all of its usual pathologies , including the tachyon ) .
the presence of tachyons is not particularly surprising , since the energy density of the theory is unbounded below .
we have not made analogous progress on the corresponding diagrams of gauge theories .
but we have studied the latter theory in the sectors with @xmath98 .
this is not particularly difficult , since the limitation on @xmath36 reduces the sum of all possible diagrams to a geometric series . nevertheless , it is interesting , for example , that the @xmath98 gluon propagator summed to all orders in perturbation theory displays no additional poles beyond that of the massless gluon itself .
in contrast the @xmath98 propagator for the fictitious 2-form field shows a bound state pole at sufficiently strong coupling @xmath99 .
this indicates that the 2-form field may be particularly important for the understanding of the strong coupling limit .
another relatively simple testing ground for our formalism is quantum field theory in low space - time dimensions , the simplest being the t hooft model ( qcd in one space and one time dimensions ) .
rozowsky and i are just finishing up a study of this model .
since this model is well understood even in the continuum limit , we used it mainly as a test of how our model approaches the continuum theory .
one interesting feature of our simultaneous discretization of @xmath1 and @xmath2 is that one can approach the continuum in different directions .
for example the approach to continuum at fixed @xmath100 is different from the approach studied in conventional dlcq .
the latter keep @xmath2 continuous throughout ( in our language this means taking @xmath73 _ first _ followed by @xmath101 ) .
we have confirmed that the same continuum physics emerges in both cases .
i would like to conclude this talk with some remarks on longer term prospects and goals .
our renewed efforts to sum planar graphs have been directly stimulated by the ads / cft duality @xcite proposed in the last couple of years .
this duality in turn was recognized @xcite to be a higher dimensional realization of t hooft s concept of holography @xcite : the vision that a consistent quantum theory of gravity requires our apparently 3 dimensional spatial world to be 2 dimensional .
i have advocated string bits as a way to realize holography in t hooft s original 2 dimensional sense : the two dimensions being the transverse dimensions of light - cone string .
however , the qcd gluons of this talk really live in 3 dimensions in spite of their description on the light - cone : the longitudinal dimension hasnt disappeared .
rather , it has been disguised as a variable newtonian mass .
( in the ads / cft duality this third dimension gets interpreted as a fifth dimension , whence holography is the statement that a 4 + 1 dimensional effective theory arises from a 3 + 1 dimensional quantum field theory . )
the defining character of string bits is that they have a _ fixed _
newtonian mass , in sharp contrast to the gluons we have been describing . to understand 3 + 1 gauge theories as part of a string bit theory , a gluon with @xmath11 must in reality be a composite of @xmath102 string bits : the gluon vertices would then be effective fission / fusion amplitudes as in nuclear physics , rather than fundamental interactions .
s. mandelstam , _ nucl .
phys . _ * b64 * ( 1973 ) 205 ; _ phys
_ * 46b * ( 1973 ) 447 ; _ nucl .
phys . _ * b69 * ( 1974 ) 77 ; see also his lectures in _ unified string theories _ , ed . m. green and d. gross ( world scientific ) 1986 . | i describe renewed efforts to establish a string description of large @xmath0 qcd by summing large `` fishnet '' diagrams
. earlier work on fishnets indicated that the usual relativistic ( zero thickness ) string theory can arise at strong t hooft coupling , at best yielding a highly idealized description , which fails to incorporate such salient features of continuum qcd as asymptotic freedom and point - like constituents .
the recently conjectured ads / cft correspondence is compatible with such limitations because it also gives a simple picture of large @xmath0 gauge theory only at strong coupling . in order to better understand how string theory could emerge from large @xmath0 qcd at strong coupling , klaus bering , joel rozowsky , and
i have developed an improved implementation of my effort of the late seventies to digitize the planar diagrams of large @xmath0 light - cone quantized qcd by discretizing both @xmath1 and @xmath2 .
this discretization allows a strong coupling limit of the sum of planar diagrams to be defined and studied .
it also provides a natural framework to explore the possible dual relationship between qcd in light - cone gauge and string theory quantized on the light - cone .
ufift - hep-00 - 9 + hep - th/0004129 1.5 cm * string from large n gauge fields + via graph summation on a @xmath1 - @xmath2 lattice * 2.cm charles b. thorn 0.5 cm _ institute for fundamental theory + department of physics , university of florida , gainesville , fl 32611 _ ( ) .5 cm |
uru@xmath2si@xmath2 is a heavy fermion compound ( space group @xmath6 ) , @xmath7 = 180 mj k@xmath8mole@xmath9 , where superconductivity ( sc ) below the superconducting transition at t@xmath10 = 1.5 k coexists with yet unidentified order parameter that is linked to a remanent antiferromagnetism ( afm ) @xcite , both appearing below t@xmath11 = 17.5 k ( for a review see @xcite ) .
the weak afm order is characterized by a propagation vector * _ q@xmath12 _ * = ( 1 0 0 ) with the ordered dipolar moments pointing along the @xmath13-axis and very small , ( 0.01 - 0.03 @xmath14 @xcite ) . in order to explain the clear anomalies in temperature dependencies of properties , notably the large entropy connected with the transition
, an unconventional phase change is postulated@xcite .
it is generally accepted that this afm order is not intrinsic but parasitic .
accordingly , this new state is called hidden order ( ho ) and is one of the most addressed topics in heavy - fermion physics research@xcite . near the ho state , different phases can be induced by external perturbations like pressure , magnetic field or substitution . a moderate pressure converts the ho phase @xcite into a static long - range antiferromagnetic ( af )
q@xmath12 order with u magnetic moments of @xmath5 0.4 @xmath14 , the so - called large moment antiferromagnetic ( lmaf ) phase . here
, the tiny remanent af order in the pure uru@xmath2si@xmath2 is usually called the small moment antiferromagnetic ( smaf ) phase .
the influence of applied magnetic field or substitutions effects are more complicated .
a strong magnetic field is necessary to suppress the ho order and generate final fermi - liquid behavior via intermediate field - induced phases between 35 and 38 t with * _ q@xmath15 _ * = ( 0.6 0 0 ) @xcite .
it has been demonstrated that few at.% substitution of various transition metals ( re , rh , fe , os , tc ) for ru destroy the sc and induce magnetic states .
the doping weakens the ho state and transform it to long - range magnetic order that is for different dopant different ( the af order with q@xmath12 = ( 1 0 0 ) being the most frequent one)@xcite . in some cases the crossover between the ho state and dopand - induced long - range magnetic order is smooth ( as in the case of fe doping ) , in other cases a gap between the two ground states exists ( the case of rh ) @xcite preceded by a tiny concentration region where the lmaf phase seems to inhomogeneously coexist . focusing on the case of rh doping , a shift of the large critical fields of metamagnetic transitions to lower fields
is found @xcite . above 4 % of rh substitution for ru
, no trace of ho state or long - range magnetic order can be discerned @xcite . however , short - range af correlations start to develop around * _ q@xmath3 _ * = ( @xmath4 @xmath4 @xmath4 ) that `` condense '' above 10 % of rh for ru into long - range af magnetic order with increasing phase transition temperatures @xcite .
the low - concentration rh region , where the ho coexists with the parasitic lmaf , has been extensively studied in the past @xcite . also the higher rh doping levels
, that exhibit magnetic order with high ordering temperatures , have been inspected to a certain extend @xcite .
however , the gap region where no ho or smaf or lmaf states are present , has been glimped using one 6 % rh crystal by burlet et al .
this region is most interesting as it should be pure fermi liquid , free of influences of any long - range order and offers a possibility to disclose the `` bare '' behavior of heavy electrons in such lightly doped rh systems .
this has motivated us to prepare a high - quality u(ru@xmath0rh@xmath1)@xmath2si@xmath2 single crystal ( see the resolution - limited rocking curve through the 1 1 0 nuclear bragg reflection in fig .
[ fig1 ] and the grown crystal in the inset of fig . [ fig2 ] ) . as we will show below
, it indeed does not exhibit any sign of smaf order or long - range order associated with * _ q@xmath12 _ * or * _ q@xmath3 _ * , respectively . no signatures of ho or sc down to 0.4 k is observed .
however , a heavy - fermion behavior remains intact , and since all the anomalies connected with ho and sc phase are removed , this `` bare '' heavy - fermion state resembles very strongly properties of the pristine uru@xmath2si@xmath2 .
nevertheless , in contrast to uru@xmath2si@xmath2 , it exhibits short - range order ( sro ) at * _ q@xmath3_*. the sro signal disappears at temperatures comparable to t@xmath11 .
our findings suggest that the heavy - fermion behavior is common to all the lightly doped uru@xmath2si@xmath2 systems and that the ho transition is a result of coherence phenomenon within the heavy - fermion liquid .
single crystalline u(ru@xmath0rh@xmath1)@xmath2si@xmath2 has been prepared by a modified czochralski method from a stoichiometric melt in high purity argon ( 6n ) atmosphere .
no weight loss was detected , allowing us to denote the nominal rh concentration as 8 % .
the maximal diameter was about 7 @xmath16 with the length of about 50 @xmath16 .
no further heat - treatment was applied after the crystal growth .
the laue x - ray backscatter images revealed very sharp reflections , along the @xmath13-axis , typical of four - fold symmetry . this method has also been used also to orient the crystal .
we cut the oriented crystal using spark erosion in three pieces of different dimension that were used either for bulk measurements or neutron diffraction .
two of them are shown in the left inset of in fig .
[ fig2 ] .
magnetization curves @xmath17(@xmath18 ) with the magnetic field applied along principal directions were measured in the temperature range between 1.8 k and 350 k using the quantum design 14 t physical properties measurements system ( ppms ) , which is part of the laboratory for magnetic measurements at hzb .
the static magnetic susceptibilities were calculated according @xmath19=@xmath17/@xmath20 , where @xmath20 denotes the field strength from @xmath17(@xmath18 ) dependencies .
magnetic measurements were extended down to 400 mk using @xmath21he fridge and a sensitive home - made ac susceptometer in a frequency range between 10 hz and 10 khz together with two magnet coil systems in order to eliminate possible artifacts originating from the detection system .
specific heat was measured on two single crystal pieces ( 12.5 and 3.5 mg , respectively ) in zero field between 1.8 k and 100 k in the same ppms using the heat capacity option in fields up to 14 t. neutron diffraction data were collected on the e1 , e2 , e4 and e5 instruments installed at the ber ii 10 mw reactor of the helmholtz - zentrum berlin . while the triple - axis instrument e1 and the normal - beam e4 diffractometer use the incident wavelength @xmath22 = 2.41 produced by the pg ( 002 ) monochromator , on the flat cone diffractometer e2 we have used @xmath22 = 1.21 produced by the ge ( 311 ) monochromator .
all the three instruments were used to characterize the quality of our single crystal and to follow and quantify the temperature evolution of the short - range signal between 1.7 k and 30 k. the e5 four - cycle diffractometer with cu - monochromator selecting the neutron wavelength @xmath22 = 0.896 was used to collect a large data set to determine the structural details .
the triple - axis instrument e1 is equipped with a single @xmath21he - detector tube and pg analyzer leading good detection rates and a possibility to filter - out inelastic processes .
we have used this instrument to separate a possible inelastic scattering contributing from the short - range signal .
the e2 flat - cone diffractometer is equipped with four two - dimensional position sensitive @xmath21he - detectors ( 300 x 300 mm@xmath8 ) .
two wide - angle rocking curves with the four detectors shifted to fill gaps between them enable an effective mapping and detection of all the diffracted signal not only within the scattering plane but also at significant distance above and below it .
the e4 diffractometer is equipped also with a two - dimensional position sensitive @xmath21he - detector .
its size of 200 x 200 mm@xmath8 makes it suitable to follow the selected signal as a function of an external parameter , in our case the temperature .
the e5 data were collected with a two - dimensional position sensitive @xmath21he - detector , 90 x 90 mm@xmath8 ( 32 x 32 pixels ) at room temperature using a single crystal with the dimensions 4 x 4 x 4 mm@xmath21 ( see the smaller single crystal shown in the left inset of in fig .
[ fig2 ] ) .
the crystal structure refinement was carried out with the program xtal @xcite .
the nuclear scattering lengths b(ru ) = 7.21 fm , b(rh ) = 5.88 fm , and b(u ) = 8.417 fm were used.@xcite for the absorption correction ( gaussian integration ) we used the absorption coefficient @xmath23 = 0.380 cm@xmath9 .
secondary extinction has been corrected using the formalism of zachariasen ( type i ) .
on e1 , e2 and e4 , we have used both , the large and smaller single crystals in the form of a truncated cone and a semi cube , respectively ( see the left inset in fig . [ fig2 ] ) .
a wide - angle rocking scans with the [ 100 ] , [ 001 ] and [ @xmath2510 ] directions vertical collected using e4 diffractometer revealed that except for additional short - range magnetic correlations present at low temperatures indexable with * _ q@xmath3 _ * = ( @xmath4 @xmath4 @xmath4 ) , all observed bragg reflections are compatible with the space group @xmath6 .
visible reflections are very sharp and resolution limited .
neither of the nuclear reflections show sizable thermal variation . as an example we show in fig .
[ fig1 ] rocking curves through the 110 nuclear reflection as recorded on e4 using the larger u(ru@xmath0rh@xmath1)@xmath2si@xmath2 single crystal at 50 k and 1.7 k. the small shift in position is caused by the thermal expansion of the .
the 100 reflection , that is for the paramagnetic space group @xmath6 forbidden and which can appear only as a consequence of an af order ( or crystal structure transformation ) has not been observed at high temperature nor at 1.7 k even after collecting the data for several hours .
the analysis based on the statistical error analysis @xcite suggests that any dipole - like moment associated with smaf order in our sample has to be smaller than @xmath5 0.008 @xmath14 .
rh@xmath1)@xmath2si@xmath2 after the extinction correction ( e5 data taken at room temperature ) .
a schematic representation of the thcr@xmath2si@xmath2 crystal structure adopted by u(ru@xmath26rh@xmath27)@xmath2si@xmath2 is shown in the right inset .
the atoms are shown as large ( red ) , intermediate ( green ) and small ( blue ) circles stand for u , ru / rh and si , respectively . the two single crystal used in the diffraction experiments
are shown in the left inset . ]
the details of the crystal structure at room temperature were determined from e5 data set that contained 664 individual reflections ( 144 unique ones ) .
it has been confirmed that u(ru@xmath0rh@xmath1)@xmath2si@xmath2 crystallizes in the thcr@xmath2si@xmath2 type of structure with the tetragonal space group @xmath6 ( no .
139 ) . in this space group
are the u , ru(rh ) , and si atoms at the wyckoff positions 2@xmath28(0,0,0 ) , 4@xmath29(0 , @xmath4 , @xmath30 ) , and 4@xmath31(0,0,@xmath32 ) , respectively .
the @xmath32 parameter for si atoms is the only free positional parameter .
the refinement of the scale factor , the positional parameter @xmath32(si ) , and the anisotropic thermal parameters of the different atoms resulted in the residuals @xmath33 = 0.030 defined as @xmath33 = @xmath34(@xmath35f@xmath36 - @xmath37f@xmath38 ) / @xmath34(@xmath37f@xmath36 ) . for the extinction parameter @xmath39 , which is related to the mosaic distribution , we obtained the value @xmath39 = 472(24 ) rad@xmath9 .
the observed versus calculated structure factors are shown in fig .
[ fig2 ] . as can be seen , a satisfactory agreement is obtained .
the numerical results of the refinements are summarized in table [ tab : table1 ] . using the lattice parameters @xmath28 = 4.0927(6 )
and @xmath13 = 9.5387(16 ) the shortest u - u neighbor are found along the @xmath28-axis with separation equal to the @xmath28-axis lattice parameter .
there are four of such neighbors .
further eight u next - nearest neighbors are found along the body diagonal at a distance of 5.5786(10 ) .
both principal lattice parameters appear to be slightly smaller than parameters of the pure uru@xmath2si@xmath2 @xcite , what is surprising in view of the effect of rh doping under which the @xmath28-axis parameter shortens and the @xmath13 parameter expands@xcite .
the most sensitive parameter of rh doping is thus a change in the @xmath13/@xmath28 ratio which should increase with increasing rh content .
indeed , this ratio is for our sample by about 0.5 % larger than for pure uru@xmath2si@xmath2 .
we therefore attribute the discrepancy in absolute values of lattice parameters to the uncertainty of the incident neutron wavelength .
however , also the positional parameter of si is slightly different from a value of 0.3710 listed in the literature for the pristine uru@xmath2si@xmath2 determined at similar conditions . here ,
however , no information on the rh doping effect is reported so far .
the calculated bond distance @xmath29(si - si ) amounts to 2.412(1 ) which is @xmath5 2 % less than for the pure uru@xmath2si@xmath2 .
let us note that we were unable to determine reliably the stoichiometry of our sample due to strong correlation among other free parameters , especially with extinction and thermal parameters .
however , we could confirm that the starting 8 % stoichiometry is compatible with our neutron crystallographic data .
.[tab : table1]crystal structure parameters of a u(ru@xmath0rh@xmath1)@xmath2si@xmath2 single crystal as determined from the e5 diffraction data at room temperature .
the thermal parameters @xmath40 ( given in 100 @xmath8 ) are in the form exp[-2@xmath41(u@xmath42 + 2u@xmath43 ) ] , where @xmath44 are indices of the relevant bragg reflection and @xmath45 and @xmath46 are reciprocal lattice constants . for symmetry reasons
the values @xmath47 , @xmath48 , and @xmath49 of the atoms u , ru(rh ) and si are equal to zero in this structure . [ cols="<,^,^,^,^,^,^ " , ] the temperature dependence of the u(ru@xmath0rh@xmath1)@xmath2si@xmath2 static magnetic susceptibility @xmath19=@xmath17/@xmath20 measured in 1 t applied along the @xmath28 and @xmath13-axis is shown in fig .
@xmath50 is very anisotropic with the @xmath28-axis magnetic susceptibility @xmath51(@xmath18 ) essentially temperature independent at a level of 0.16 10@xmath52 m@xmath21mol@xmath9 in the whole range measured . for this direction
no analysis according to a curie - weiss ( cw ) law @xmath53(@xmath18 ) = @xmath54/(@xmath18 - @xmath55 ) , where @xmath54 denotes the curie constant and @xmath55 the paramagnetic curie temperature , is possible . for comparison , the @xmath53(@xmath18 ) follows a cw law only in the high - temperature limit . above 300 k , the best fit shown in the left inset of fig .
[ fig3 ] by the solid line through experimental @xmath53(@xmath18 ) points yields @xmath54 = 1.9316 ( 5 ) 10@xmath56 and a negative @xmath55 = -20.6 ( 8) k. the effective magnetic moment per uranium atom of 3.51 ( 1 ) @xmath14 derived from the curie constant is close to both , the effective moment found for the pure uru@xmath2si@xmath2 system along the @xmath13-axis and the effective moment expected for a localized u@xmath57 or u@xmath58 configurations ( 3.58 and 3.62 @xmath14 , respectively ) .
= @xmath17/@xmath20 of u(ru@xmath0rh@xmath1)@xmath2si@xmath2 single crystal measured in 1 t applied along the principal axes . in the left inset
we show the temperature dependence of the inverse magnetic susceptibility measured in 1 t applied along the tetragonal axis together with the best fit to a curie - weiss law to data above 300 k. in the right inset the low - temperature detail of the @xmath13-axis magnetic susceptibilities measured in 1 and 14 t with field applied along the tetragonal axis in zero - field cooled and field cooled regimes is shown .
note that there is no sharp shoulder around 17 k as in the case of the pristine uru@xmath2si@xmath2 . ]
rh@xmath1)@xmath2si@xmath2 single crystal magnetization measured at various temperatures between 1.8 k and 60 k with field applied along the @xmath28 and the @xmath13-axis .
note the nearly linear @xmath59 behavior throughout . ] as the temperature is lowered , the @xmath53(@xmath18 ) deviated progressively from a cw behavior , exhibits an maximum around 46 k and decreases strongly at lower temperatures .
it has a weak smooth inflection point around 15 k and saturates eventually around a value of 0.85 10@xmath52 m@xmath21mol@xmath9 in the low - temperature limit .
this value is six times larger than the value found for @xmath51 at low temperatures and by @xmath5 40 % larger than @xmath53 at low temperatures observed for pristine uru@xmath2si@xmath2 .
the magnetic susceptibility along the @xmath13-axis , @xmath53(@xmath18 ) is in the high - temperature limit independent of the applied field . at lower temperatures , however , differences are visible .
first , the temperature at which the @xmath53(@xmath18 ) exhibits maximum shifts slightly to lower values . at 14 t
it is found around 43 k. second , @xmath53(@xmath18 ) shows slight field dependence .
its values measured at 14 t are generally lower above 60 k and larger below this temperature with respect to 1 t values . at lower fields ,
there is also a history dependence manifested by a tiny splitting of zero filed cooled ( zfc ) and field cooled ( fc ) @xmath53(@xmath18 ) curves measured at 1 t that merge again around 10 k. with increasing field the difference between the fc and zfc curves vanishes .
these observations are documented in the right inset of fig .
[ fig3 ] .
ac susceptibility measurements down to 400 mk performed in a wide frequency range did not revealed any anomalies that would suggest any ho , magnetic or superconducting phase transition .
so , although the overall temperature dependence of both , @xmath51(@xmath18 ) and @xmath53(@xmath18 ) resembles very strongly the behavior of uru@xmath2si@xmath2 , no phase transitions down to 400 mk can be detected in the case of u(ru@xmath0rh@xmath1)@xmath2si@xmath2 . in fig .
[ fig4 ] we show the field dependence of the magnetization measured along the @xmath13-axis ( @xmath28-axis ) between 1.8 k and 60 k ( 50 k ) in fields up to 14 t. clearly , the response along the @xmath13-axis is much larger than along the @xmath28-axis .
the response to the magnetic field along the @xmath28-axis is nearly temperature independent in the measured range , whereas for the @xmath13-axis it is nearly a factor of two stronger than at 1.8 k. these findings are in agreement with magnetic susceptibility results and establish a huge magnetocrystalline ising - like anisotropy in u(ru@xmath0rh@xmath1)@xmath2si@xmath2 similar to uru@xmath2si@xmath2 .
absence of any anomalies up to 14 t along both directions suggest that u(ru@xmath26rh@xmath27)@xmath2si@xmath2 remains paramagnetic even at 14 t. however , we note that the magnetization does not increase strictly linearly with field .
there is small but finite concave shape of magnetization curves up to 50 k ( compare in fig .
[ fig4 ] for instance magnetization recorded at 40 k and 50 k with curve at 60 k ) . for the @xmath28-axis direction is the response linear at all temperatures .
rh@xmath1)@xmath2si@xmath2 electrical resistivity for current along the @xmath28 and the @xmath13-axis together with the best fit described in the main text . in the inset
the electrical resistivity measured along the @xmath28-axis and the @xmath13-axis in the whole temperature range in zero field and 14 t applied along the tetragonal axis is shown . ] in the inset of fig .
[ fig5 ] we show the electrical resistivity for current along the @xmath28 and the @xmath13-axis measured in zero and 14 t external field applied along the tetragonal axis in the temperature range between 2 and 350 k. @xmath60 is along both directions rather large at high temperatures and very anisotropic in the whole temperature range . at 350 k , the electrical resistivity measured along
the @xmath13-axis , @xmath61 , reaches @xmath5 240 @xmath62cm@xmath9 , while the @xmath28-axis value @xmath63 is about 40 % larger . both , @xmath61(@xmath18 ) and @xmath63(@xmath18 ) increase upon lowering the temperature indicating a formation of a heavy - fermion state , exhibit a maximum at slightly different temperatures and decrease strongly below 50 k to level - off in the low - temperature limit . no anomalies indicating phase transitions are visible .
while @xmath63(@xmath18 ) shows a maximum at 55 k , @xmath61(@xmath18 ) peaks at higher temperature of about 75 k. the ratio @xmath63/@xmath61 increases upon lowering the temperature very slightly and attains a maximum around 55 k. the high - temperature anisotropy is smaller than the anisotropy found for uru@xmath2si@xmath2 @xcite ( where @xmath63 is about twice as large as @xmath61 ) , however , it remains in u(ru@xmath0rh@xmath1)@xmath2si@xmath2 finite also in the low - temperature limit , where the anisotropy for the pristine compound vanishes .
the residual resistivity is large most probably due to the disorder caused by the rh substitution . the low - temperature parts with current along both directions , shown in the main panel of fig .
[ fig5 ] , can not be described by perfect fermi - liquid dependence of the form @xmath64(@xmath18 ) = @xmath65 + @xmath66 , @xmath67 = 2.0 .
the best fit to data between 2 and 10 k yields @xmath67 @xmath5 2.2 for both directions .
even better agreement with data in the same temperature range is obtained for expression @xmath64(@xmath18 ) = @xmath65 + @xmath68 + @xmath69(1 + 2@xmath18/@xmath70)e@xmath71 .
this formula has been introduced in order to account for the influence of an energy gap @xmath70 in the dispersion relation of magnetic excitations caused by strong electron - magnon coupling @xcite and was also used in order to describe the electrical resistivity behavior of pure uru@xmath2si@xmath2 @xcite .
the best fits to this formula between 2 k and 10 k are shown in fig .
[ fig5 ] by solid lines through the experimental points .
the smooth @xmath64(@xmath18 ) behavior eliminates any possibility of ho or sc phase transitions . for numerical results see appendix .
the effects of the external field on the electrical resistivity of u(ru@xmath0rh@xmath1)@xmath2si@xmath2 are very anisotropic .
while the application of field along the hard magnetization @xmath28-axis leads to identical temperature dependencies , for the field applied along the @xmath13-axis we observe readily observable changes .
this is displayed in the inset of fig .
small shifts of temperatures where maxima in @xmath64(@xmath18 ) occur are detected . at 14 t the @xmath63(@xmath18 ) and @xmath61(@xmath18 ) maxima shift to about 60 and 73 k , respectively .
the field increases ( decreases ) the resistivity at temperatures below ( above ) 30 - 35 k along both directions .
this leads to a magnetoresistance of opposite signs below and above this temperature .
( @xmath72-@xmath73)/@xmath72 is shown in fig .
[ fig6 ] . for the 14 t it reaches 18 % around 8 - 10 k for both directions . the detailed temperature dependence of @xmath64(@xmath18 ) in magnetic fields
is also modified . in 14 t (
in contrast to zero field ) it can be , however , described by @xmath64(@xmath18 ) = @xmath65 + @xmath68 dependence suggesting a full restoration of the heavy fermi liquid behavior . for numerical results
see appendix . while there is nearly no field dependence along the @xmath28-axis , the electrical resistivity is strongly affected by a field applied along the @xmath13-axis . in fig .
[ fig7 ] we show the magnetoresistance of the u(ru@xmath0rh@xmath1)@xmath2si@xmath2 single crystal measured with the current along the @xmath28 and @xmath13 axes at various temperatures as a function of field applied along the @xmath13-axis . except for 10 and 15 k ( shown in the inset of fig .
[ fig6]a ) there is no hysteresis between field sweeps up and down only increasing field curves are shown .
hysteresis is found for both current orientations only between 10 and 15 k. rh@xmath1)@xmath2si@xmath2 single crystal measured with the current along the @xmath28-axis ( a ) and the @xmath13-axis ( b ) at various temperatures as a function of field applied along the @xmath13-axis .
full lines represent best fits to expression ( @xmath74 - @xmath72)/@xmath72 = @xmath75 + @xmath76 . in the inset of panel ( b )
we show the temperature dependence of the fit parameters to this formula for both electrical current orientations . in the inset of panel ( a ) we document a hysteretic behavior for current along the @xmath28-axis that is found for both current orientations only at 10 and 15 k. ] at low temperatures the electrical resistivity increases with the applied field @xmath77 = @xmath78 and decreases at temperatures above 30 k ( see also fig .
[ fig6 ] ) for both current orientations . at low temperatures the field dependence can not be described quadratically in the form ( @xmath74 - @xmath72)/@xmath72 = @xmath79 , @xmath67 = 2.0 .
this is in contrast to the pure uru@xmath2si@xmath2 that is reported to exhibit the quadratic dependence @xcite . instead
, our sample s magnetoresistance can be fit by an expression ( @xmath74 - @xmath72)/@xmath72 = @xmath75 + @xmath76 .
the temperature dependencies of fit parameters from best fits are shown in the inset of of fig .
the `` a '' parameter ( linear in field ) is negative below 15 k and positive at all temperatures above this temperature .
the `` b '' ( quadratic term ) is at first positive , increases with increasing temperature , peaks around 15 k and decreases above this temperature .
it changes its sign , reaches large negative value around 40 k and decreases in absolute value with further increasing temperatures .
both parameters exhibit qualitatively the same behavior for the two current orientations and it can be seen that at temperatures above @xmath5 70 - 80 k the linear term diminishes and the `` c''-axis field dependence according to ( @xmath74 - @xmath72)/@xmath72 = @xmath79 with @xmath67 = 2.0 is restored .
rh@xmath1)@xmath2si@xmath2 specific heat together with the best fit containing an exponential term ( see the main text ) . in the inset
the low - temperature detail of the @xmath80 @xmath81 @xmath82 dependence is shown along with the best fits to a linear ( solid , blue line ) and power - law ( dashed , red line ) dependencies , respectively . ]
the temperature dependence of the specific heat down to 1.8 k is shown in fig .
[ fig8 ] . no signature of any phase transition can be discerned in the whole temperature range .
this observation is in accord with the magnetic and electrical resistivity bulk measurements and document clearly the absence of the hidden order .
the specific heat @xmath54 can be best fit between 1.8 k and 30 k to a formula @xmath54 = @xmath83 + @xmath84 + @xmath85exp@xmath71 , where @xmath7 denotes the electronic low - temperature specific heat coefficient , @xmath86 relates to the debye temperature @xmath87 via expression @xmath88 = 12*@xmath89/5@xmath86 and @xmath70 denotes an energy gap in the putative dispersion relation of magnetic excitations @xcite .
the best - fit parameters to this formula yields @xmath7 = 111.9 ( 3.9 ) mj /(mol
k@xmath8 ) , @xmath87 = 181(1 ) k and @xmath70 = 13.6 ( 7 ) k. the gap value is significantly lower than the value of 115 k found for the pure uru@xmath2si@xmath2 @xcite and again suggests the absence of ho and sc in the present system . let us note that the expression that neglects the gap term deviates from experimental data between 2 and 15 k. however , the deviation is small ( less than 1 % ) and , at maximum , amounts to @xmath5 15 mj /(mol k ) .
the difference between the `` non - magnetic background '' approximated by the formula without the gap term and measured specific heat can be interpreted as a contribution due to fluctuating magnetic moments .
magnetic entropy obtained by integration of this difference divided by temperature up to 15 k is tiny and amounts at most to 4 * 10@xmath902 . in the inset of fig .
[ fig8 ] we show the low - temperature detail of the @xmath80 @xmath81 @xmath82 dependence .
as can be seen , experimental data show progressively downward curvature at the lowest temperatures .
this is also a reason why it is not possible to deduce the low - temperature specific heat coefficient using the usually used expression @xmath80 = @xmath7 + @xmath91 .
however , above @xmath5 17 k the data follow reasonably such a dependence .
the best fit to data between 17 and 27 k yields @xmath7 = 125.6(1.6 ) mj /(mol k@xmath8 ) and @xmath87 = 177(2 ) k. this fit is shown in the inset of fig .
[ fig8 ] by the full ( blue ) line .
the curvature of the data below 15 k can be accounted for by a power law in the form @xmath80 = @xmath7 + @xmath92 .
the best to this expression between the lowest temperature and 14.5 k yields a lower value of @xmath7 = 92.2(1.7 ) mj /(mol k@xmath8 ) and n = 0.40(2 ) k. rh@xmath1)@xmath2si@xmath2 specific heat measured at different magnetic fields applied along the tetragonal axis . in the inset
the magnetic entropy calculated by integration of data shown in the main panel ( from which an estimate of a non - magnetic analog contribution has been subtracted ) divided by the temperature up to 14 k is shown . ] in fig .
[ fig9 ] we show the low - t detail of the field dependence of the specific heat as a function of temperature measured at various magnetic fields applied along the @xmath13-axis
. such appears below 15 k @xmath93 and increases with increasing field in a non - linear fashion . at constant temperature
the field dependence can be described by a power law according to @xmath94 + @xmath95 with @xmath67 dependent on the field . @xmath67 is close to 1.90 at 2 k and decreases slightly with increasing temperature . at 7
k it amounts to 1.7 . above this temperature
the parameter @xmath67 decreases faster and approaches zero around 14 k. at higher temperatures the specific heat seems to decrease marginally with increasing field .
apparently , the magnetic entropy @xmath96 calculated by integration of the specific heat from which the `` nonmagnetic '' background has been subtracted and divided by temperature between the lowest temperature and 14 k at particular field increases with increasing field ( see the inset of fig .
the dependence is again non - linear . at low fields
@xmath96 quadratically increase as a function of the applied field , above @xmath5 4 t the dependence changes to a linear one . at 14 t
the magnetic entropy amounts to 0.047 @xmath97 per formula unit .
rh@xmath1)@xmath2si@xmath2 single crystal measured at 1.8 k ( a ) and at 20 k ( b ) in the ( @xmath98 ) orientation using e2 diffractometer transformed into the reciprocal space .
note that this instrument senses a significant portion of the reciprocal space also above and below the scattering plane .
sharp strong nuclear bragg reflections of the crystal main phase are observed along with broader features indexable with * _ q@xmath3 _ * = ( @xmath4 @xmath4 @xmath4 ) . *
_ q@xmath3 _ * and -*_q@xmath3 _ * are shown in ( a ) by arrows . ]
= @xmath4 @xmath99 = @xmath4 @xmath100 = @xmath4 position at 1.8 k and at 25 k as measured on the e4 diffractometer .
( a ) reciprocal scan along the [ @xmath101 @xmath101 0 ] direction .
the scan along the [ 0 0 @xmath100 ] direction is shown in panel ( b ) .
full line through the experimental points are best fits to a lorentzian profile .
the short black line ( `` h '' ) at the bottom shows the instrument resolution at the current position . ] in fig . [
fig10](a ) and ( b ) we show the reciprocal ( @xmath98 ) plane of the u(ru@xmath0rh@xmath1)@xmath2si@xmath2 single crystal as measured using neutron diffraction technique on the e2 instrument at 1.8 k and 20 k , respectively .
as can be seen , at 2 k , apart from nuclear bragg reflections , we also observe an additional , weaker and broader , signal at places that are indexable with * _ q@xmath3 _ * = ( @xmath4 @xmath4 @xmath4 ) .
this signal has clearly lorentzian profile and at low temperatures its full width at half maximum is three to four times larger than the width of the nuclear reflections that have gaussian profiles .
this fact is documented in fig . [ fig11](a ) and ( b ) ( please , compare with the 110 nuclear reflection shown in fig .
[ fig1 ] ) , where we show representative reciprocal scans along [ @xmath101 @xmath101 0 ] and [ 0 0 @xmath100 ] directions through the ( @xmath4 @xmath4 @xmath4 ) reciprocal space position as measured on the e4 diffractometer .
both scans can be fit to a lorentzian profiles with the full width at a half maximum being about four times larger than the resolution .
the correlation lengths are found to be @xmath5 100 perpendicular to the @xmath13-axis and @xmath5 200 along the tetragonal axis .
this indicates that this magnetic signal is not due to a long - range magnetic order but short - range order ( sro ) .
such sro has been previously observed in u(ru@xmath26rh@xmath27)@xmath2si@xmath2 single - crystalline sample with @xmath103 = 0.06 @xcite .
however , correlation lengths in our system are significantly larger .
this agrees with the general tendency towards afm order with increasing doping of rh for ru . with increasing rh content beyond our 8 % rh crystal ,
the magnetically ordered phase appear and the transition temperature increase@xcite . in our system
, however , no phase transition down to 0.4 k has been detected .
= @xmath4 @xmath99 = @xmath4 @xmath100 = @xmath4 reciprocal space position between 1.7 k and 21 k. in the inset we show an energy scan through this position using the tripple - axis spectrometer e1 at 1.7 k and 21 k. ] as the temperature is increased , the sro intensity decreases .
no traces of sro could be observed neither in the e2 nor in the e1 or e4 data taken above 20 - 25 k. the temperature dependence of the peak intensity at the ( @xmath4 @xmath4 @xmath4 ) position as measured on e1 and e4 is shown in fig .
[ fig12 ] .
as can be seen , it decreases with increasing temperature and exhibit an inflection point around 7 - 8 k with a long tail at higher temperatures .
the background level is reached at lower temperatures on e1 as compared to e4 data .
nuclear bragg reflections are shown in the same temperature range without a broadening or temperature dependencies .
as the e4 instrument detects all the scattered neutrons irrespective of their energy , this observation suggests that some portion of detected neutrons at this reciprocal space position undergo inelastic scattering process .
this is documented in the inset of fig .
[ fig12 ] which show the constant- * _ q _ * energy scan at the ( @xmath4 @xmath4 @xmath4 ) position at 1.7 k and at 21 k recorded using the e1 spectrometer .
it is apparent that the signal seen at the lower temperature is present within the whole energy window of the e1 spectrometer .
nuclear bragg reflections , on the other hand , are much narrower in the energy and fit into this window entirely .
this suggests that the magnetic signal is quasielastic in nature but extends most probably also outside the e1 energy window .
this intensity , not detected at e1 , is however collected on e4 , thus causing the difference in the temperature dependences seen using the two instruments .
based upon our detailed and systematic bulk measurements of single crystal u(ru@xmath0rh@xmath1)@xmath2si@xmath2 we have verified its heavy fermi liquid behavior over the entire temperature and field ranges .
however and most important , our 8 % rh substituted crystal does not exhibit any sort of phase transition or long - range order down to 0.4 k. instead , according to our neutron scattering experiments , it displays short - range order of the * _ q@xmath3 _ * = ( @xmath4 @xmath4 @xmath4 ) wave vector with a correlation length of 100 (in plane ) and 200 (along the @xmath13-axis ) . the * _ q@xmath3 _ * propagation vector is distinctly different from the * _ q@xmath12 _ * = ( 1 0 0 ) characterizing the parasitic smaf @xcite ( fe or os doping or pressure induced lmaf ) or q@xmath15 = ( 0.6 0 0 ) propagation vector of the very high field spin density wave ( sdw ) observed in pure uru@xmath2si@xmath2 @xcite . the long - range character of the * _ q@xmath3 _ * af order appears at slightly higher rh concentrations @xcite , not reached in our 8 % rh - doped crystal .
thus we have destroyed the ho and superconductivity with rh substitution allowing a pure heavy fermi liquid to remain .
this then begs the question : how does this rather dilute rh substitution ( one extra 4@xmath29 electron ) affect the band structure and fermi surface ?
u(ru@xmath0rh@xmath1)@xmath2si@xmath2 is similar to the heavy fermion behavior of uru@xmath2si@xmath2 in its field dependences at least up to our 14 t limit .
strong ising- like anisotropy is formed along the @xmath13-axis with little or no field effects in the plane .
since there are no orderings in our crystal , the applied field only quantitatively induces small changes in the magnetization , specific heat and resistivity .
our maximum field of 14 t is insufficient to create the new magnetic state ( * _ q@xmath15 _ * ) that we expect to appear above 22 t. here such high field magnetization and neutron experiments are planned .
the interesting observation of q@xmath3 short - range order that is absent in undoped uru@xmath2si@xmath2 ( where parasitic sro with * _ q@xmath12 _ * is found ) must be further studied .
our preliminary indications of magnetic fluctuations in the inelastic neutron scattering point to a precursor of the higher rh concentration magnetic order found in ref .
finally we note the difference between rh - tuning ( leading to magnetic order with * _ q@xmath3 _ * ) , pressure tuning ( magnetic order with * _ q@xmath12 _ * ) and field tuning ( characterized by * _ q@xmath15 _ * ) of the ho behavior of uru@xmath2si@xmath2 towards lmaf : an intriguing problem for theoretical consideration .
as the temperature dependence of the electrical resistivity deviates from the quadratic dependence , we have utilized formula @xcite @xmath64(@xmath18 ) = @xmath65 + @xmath68 + @xmath69(1 + 2@xmath18/@xmath70)e@xmath71 that was used previously for the pristine uru@xmath2si@xmath2 @xcite .
these fits provide an excellent description of the temperature dependences both , along the @xmath28 and @xmath13 axis .
the best fits between 2 and 10 k yield @xmath104 = 95.7(2 ) @xmath62 cm , @xmath105 = 0.35(5 ) @xmath62cmk@xmath106 and @xmath107 = 7.7(1.3 ) k for the @xmath28-axis direction and @xmath108 = 66.8(1 ) @xmath62 cm , @xmath109 = 0.25(2 ) @xmath62cmk@xmath106 and @xmath110 = 10.7(1.8 ) k for the @xmath13-axis direction , respectively .
they are shown in fig .
[ fig5 ] by solid lines through the experimental points .
values of @xmath70 parameters are for our sample significantly lower than for pristine uru@xmath2si@xmath2 @xcite .
the field dependence of the electrical resistivity in 14 t applied along the @xmath13-axis is of the @xmath64(@xmath18 ) = @xmath65 + @xmath68 type .
the best fit to this dependence between 2 and 10 k yields @xmath111 = 106.4(3 ) @xmath62 cm and @xmath112 = 0.713(5 ) @xmath62cmk@xmath106 for the @xmath28-axis direction and @xmath113 = 72.9(1 ) @xmath62 cm and @xmath114 = 0.442(3 ) @xmath62cmk@xmath106 for the @xmath13-axis direction , respectively . | we report the crystal structure and highly - anisotropic magnetic , transport and thermal properties of an exceptionally good single crystal of u(ru@xmath0rh@xmath1)@xmath2si@xmath2 prepared using a modified czochralski method .
our study , that also includes neutron diffraction results , shows all the heavy - fermion signatures of pristine uru@xmath2si@xmath2 , however , the superconductivity , hidden order and remanent weak antiferromagnetic orders are absent . instead , the ground state of the doped system can be classified as a spin liquid that preserves the heavy - fermion character .
u(ru@xmath0rh@xmath1)@xmath2si@xmath2 exhibits a short - range magnetic order distinguished by reflections of a lorentzian profile at * _ q@xmath3 _ * = ( @xmath4 @xmath4 @xmath4 ) positions that disappear above @xmath5 15 k. the short - range order seems to be a precursor of a long - range magnetic order that occurs with higher rh concentration . we indicate that these short - range fluctuations involve , at least partially , inelastic scattering processes . |
diffusion is always related to noise in some way . in the case of brownian motion , and in most studies of diffusion ,
the noise is thermal in origin .
thermal noise is , in principle , generated by a deterministic many - particle system , but noise can also be generated by smaller deterministic systems if they possess the right dynamical properties . in the case of diffusion of large molecules adsorbed on a surface , the internal dynamics of the molecules can be such a source of noise .
experimental @xcite and numerical @xcite evidence exists demonstrating that rotational and other internal degrees of freedom affect the diffusion . in some cases clusters of molecules have been shown to diffuse more rapidly than single particles @xcite .
all this suggests that the internal degrees of freedom can play an important role .
they may act as a finite heat bath and drive the diffusion . for this to happen
, two important conditions must be met .
first of all , the internal dynamics must be chaotic , so as to produce noise .
secondly , the internal degrees of freedom must be suitably coupled to the motion on the substrate , so that the noise can be transferred from one system to the other .
a large body of work exists on the relationship between noise and chaos in finite systems , see for instance references @xcite . in reference @xcite it was shown that in a formally similar system , the bouncing - ball billiard , such a relation between internal noise and global diffusion can indeed be made .
under certain conditions , the system exhibits a separation of time scales between slow global coordinates and the fast `` internal '' dynamics of the particle .
if the fast internal dynamics are chaotic it can be shown that they act as a source of noise for the slow global coordinates and lead to diffusion , the size of which can be estimated .
if the internal dynamics are not chaotic , there is no noise , and hence no diffusion . in this paper
we study the relation between time - scale separation , fast chaos , and diffusion of molecules on a substrate .
we describe the molecules and their interaction with the substrate with realistic , classical , atomistic hamiltonians and show how even pair - wise harmonic potentials can lead to nonlinearities and chaotic behaviour .
we show that in benzene chaotic internal degrees of freedom do indeed lead to diffusion of the molecule on a graphite substrate , without any addition of thermal noise . in the following section [ sec : tss ] we first discuss time - scale separation in adsorbed molecules , and review the case of the bouncing - ball billiard . in section [ sec : sourcesnlin ]
we explain how the internal degrees of freedom of a molecule can produce noise through nonlinear dynamics , which conditions must be met , and which properties of the molecule may enhance the chaos . finally , in section [ sec : benzene ] , we discuss a simple atomistic model for a benzene molecule , adsorbed on a graphite surface , and show some preliminary results , confirming that chaos in the internal degrees of freedom does indeed affect the diffusion on the substrate .
when a molecule is adsorbed weakly on a surface , i. e. it is not strongly distorted by the surface , the internal forces are stronger than the forces exerted by the substrate , and so the internal coordinates change more rapidly than the coordinates of the centre of mass on the substrate .
the system has two separate time scales , a long one for the motion of the centre of mass on the substrate and a shorter one for the internal degrees of freedom of the molecule .
more specifically , the equations of motion can be split into those for the slow motion of the centre of mass , described by @xmath0 ( containing both position and momentum ) , and those for the fast internal coordinates @xmath1 , @xmath2 with @xmath3 and @xmath4 both well - behaved functions of order @xmath5 , and the time - scale separation parameter @xmath6 a small number .
time - scale separation is very common in nature and exists on many different scales and levels of complexity , from the fairly simple orbits of planets and moons to the extremely complicated interaction between the day - to - day weather and the climate . in general , to simplify description of the slow subsystem , one wishes to eliminate the fast variables . in practice , however , the fast variables can not simply be removed , but must be replaced with some effective behaviour .
several strategies exist for dealing with systems with multiple time scales . if the fast subsystem behaves ( quasi-)periodically , the coupling to the slow system can be averaged out in the treatment of the slow system@xcite .
the coupling is replaced by an effective deterministic coupling term . if the fast subsystem is highly disordered and has a very large number of degrees of freedom , it can be approximated by an infinite heat bath @xcite .
a fast subsystem decays to equilibrium quickly , so that the coupling which affects the slow system can be approximated by equilibrium behaviour . using projection - operator methods , such as described in van kampen @xcite or in fick & sauermann @xcite
, it can be shown that , in general , coupling to a system consisting of many fast degrees of freedom can be approximated by stochastic forces and damping . here
, we wish to consider fast subsystems which can are neither infinite nor ( quasi-)periodic , but can be seen as finite heat baths .
it has been suggested that chaotic fast degrees of freedom can not be distinguished from noise @xcite .
therefore , we would like to replace the fast degrees of freedom by some suitably chosen noise .
if the fast system is sufficiently well - behaved and correlation in it decays exponentially , the noise should be gaussian white noise . in references
@xcite it was shown through the use of projection - operator methods , that , under certain conditions , a reduction similar to the one for large systems is also valid if the slow system is coupled to a small number of fast chaotic degrees of freedom .
similar results were also obtained for different conditions in references @xcite through the use of the zwanzig projection method , and in reference @xcite . in these
works the validity of linear response ( or time - scale separation ) is considered carefully . in all of these results ,
the fast chaos acts as a finite thermodynamic heat bath and leads to stochastic driving and damping of the slow system , with clear signatures @xcite of the finite size of the fast system , and the finite amount of energy stored in it .
when a dynamical system explores the entire phase space for almost all initial conditions , it is called mixing . if the fast subsystem has this property , the system s long - time behaviour is independent of the initial conditions .
if the fast subsystem also has exponential decay of correlation , then the slow coordinates can be described by a fokker - planck equation for their probability density . in reference
@xcite explicit expressions were derived for the effective drift and diffusion in terms of averages over trajectories of the fast subsystem for fixed slow coordinates .
these expressions can be used to calculate the effective behaviour of the slow subsystem @xcite . due to the mixing , the expressions for the diffusion and drift are uniquely defined , and can not depend on the history of the slow subsystem .
however , mixing is not strictly necessary . a somewhat relaxed but more complicated condition , that only the chaotic region of the fast subsystem be mixing and deformed continuously with the slow coordinates , suffices @xcite .
this property is much more common in physically motivated ( hamiltonian ) dynamical systems with a small number of degrees of freedom than mixing itself . in order to gain understanding of this kind of time - scale separation in physical systems , in reference
@xcite de wijn and kantz studied a simpler but still physical discrete dynamical system called the bouncing - ball billiard .
it is a simple model for transport and is formally similar to transporting systems with internal degrees of freedom , such as molecules adsorbed at surfaces .
the bouncing ball billiard is a two - dimensional extension of the one - dimensional bouncing - ball problem , which has been studied in great detail both experimentally @xcite and theoretically @xcite .
it was first introduced by mtys and klages in reference @xcite .
as shown in figure [ fig : bbb ] , a point particle subject to a gravitational field @xmath7 bounces inelastically ( restitution coefficient @xmath8 parallel to the surface and @xmath9 perpendicular to it ) on a vibrating plate ( frequency @xmath10 and amplitude @xmath11 ) .
the vibrating plate has a symmetric tent shape , with small slope @xmath12 .
the vertical motion of the point particle corresponds to the internal degree of freedom of the molecule , while the horizontal motion on the plate with a periodic shape corresponds to the motion of the centre of mass on the periodic substrate .
note that the `` internal '' degrees of freedom of the bouncing ball are not hamiltonian .
however , what is more relevant for the issues of effective diffusion is the structure of the phase space : as is the case for hamiltonian systems with a small number of degrees of freedom the system can exhibit both chaotic and periodic behaviour ( is not mixing ) .
one - dimensional hamiltonian systems , on the other hand , are never chaotic . for small slope ,
the bouncing - ball billiard has a time - scale separation between the horizontal and vertical motion .
this implies that if the vertical motion is chaotic , the horizontal motion should be diffusive , while if the vertical motion is periodic , the horizontal motion should be deterministic .
this was confirmed using molecular - dynamics simulations . due to the relative simplicity of the dynamics
, the effective diffusion and drift can be evaluated ( see equations ( 18)(23 ) of reference @xcite ) . when compared to the numerical simulations ,
the results show that , in this model system , the time - scale separation theory for finite chaotic fast systems can be used to successfully predict the diffusive behaviour and approximate diffusion and drift .
through a separation of time scales , chaos in the internal degrees of freedom of a molecule can lead to diffusion on a substrate in the same way as the vertical chaos in the bouncing - ball billiard leads to diffusion in the horizontal profile of the driving plate . in order for the molecule to generate noise and for this to contribute to the diffusion , the internal dynamics must be chaotic even without external fields . when a system is chaotic , it is exponentially sensitive to initial conditions .
the rates at which infinitesimal perturbations of the initial conditions blow up for long times are referred to as the lyapunov exponents .
given the finite resolution of any observation , a trajectory with such a sensitivity becomes unpredictable for long times .
seemingly random noise is thus generated deterministically from the infinite detail of the initial conditions .
the chaotic hypothesis , put forward by gallavotti and cohen @xcite states that many - particle systems , such as the ones typically studied in statistical physics , are strongly chaotic . for systems with fewer degrees of freedom , such as the molecules studied in this paper
, this assumption can not be made .
we must first investigate how and when chaos appears in such systems .
only nonlinear dynamical systems may be chaotic @xcite .
in general , the nature of a trajectory in a nonlinear system depends on the initial conditions .
it is not unusual for both regular and chaotic behaviour to occur in the same system , but in different parts of the phase space . in principle , molecules should be treated quantum mechanically .
the study of chaos in quantum mechanical systems is a field of its own . for more details ,
see , for instance , reference @xcite . in general
, it is possible to treat a quantum mechanical dynamical system classically if the energy spacings are sufficiently small compared to the total amount of energy available to excite states . here , we use realistic , classical potentials to describe molecules .
many such potentials exist in the literature , but here we choose to start from the much used expressions of the tripos 5.2 force field @xcite . in this description ,
the bending and stretching vibrations around equilibrium positions contribute purely quadratic terms to the potential .
most atomistic descriptions are similar , in the sense that the pair - wise potentials for bond length and angles are nearly or completely harmonic . naively , one would expect that pair - wise harmonic potentials can not lead to nonlinear dynamics , but , as we will demonstrate below , the geometry of a molecule can still lead to nonlinear behaviour .
the model systems used here are all hamiltonian and therefore conserve energy , as well as phase space volume and several other quantities . conserved quantities and symmetries of the dynamics
can be used to generate perturbations which do not grow or shrink exponentially ( lyapunov exponent zero ) , and therefore do not contribute to chaos .
perturbations in the centre of mass coordinates , for instance , grow at most linearly , not exponentially . due to the symmetries and conserved quantities , there needs to be a certain minimum number of coupled degrees of freedom before a hamiltonian system can display chaos . a one - dimensional dimer without external fields ,
for instance , can not be chaotic , as it has four - dimensional phase space , and also four zero modes .
three zero modes are associated with perturbations in the position and momentum of the centre of mass , and time - translation invariance . due to the conservation of phase - space volume
, the fourth lyapunov exponent must also be zero . in three dimensions ,
a dimer has a 12-dimensional phase space of which the six components related to the centre of mass decouple , and can not produce chaos .
the rotational symmetry generates four perturbations that can not grow .
finally , there is time - translation invariance , combined with the conservation of phase - space volume .
this leaves no degrees of freedom for the production of chaos . for the possibility of chaos to exist , therefore
, the molecule must consist of at least three particles .
the simplest three - particle system is a trimer , such as displayed in figure [ fig : trimer ] . for simplicity , we take one of the particles at the end of the trimer to be much heavier than the other two and this particle s position and orientation are kept fixed .
the two lighter atoms have masses @xmath13 and @xmath14 respectively , while the bond lengths are @xmath15 and @xmath16 .
the angles @xmath8 and @xmath17 finally describe the geometry of the bonds . in the equilibrium configuration @xmath15 , @xmath16 , @xmath17 , and @xmath8
take the values @xmath18 , @xmath19 , @xmath20 , and @xmath21 respectively , at the global energy minimum . without loss of generality
, @xmath21 can be taken to be @xmath22 .
as mentioned above , we use a pair - wise harmonic potential for the bending and stretching .
the typical frequencies associated with the bending are lower than those associated with the stretching , see for instance the constants in reference @xcite . with the low frequencies come relatively large amplitudes for the same amount of energy , and so nonlinear terms in the angles or their time derivatives are larger than those in the bond lengths or their derivatives . for the purpose of demonstrating the existence of large nonlinear terms in the dynamics , it therefore suffices to keep the bond lengths fixed .
let @xmath23 and @xmath24 be the spring constants of the harmonic terms for @xmath17 and @xmath8 respectively .
the total potential energy can be written as @xmath25 the kinetic energy expressed in these coordinates is more complicated , @xmath26 \right)^2 \right.\nonumber\\ & & \phantom{[}\left.\null + \left(\frac{d}{dt}[r \cos\alpha + r \cos(\pi-\phi+\alpha)]\right)\right]~,\\ & = & \frac12 m r^2 \dot\alpha^2 + \frac12 m \left [ ( r^2+r^2 ) \dot\alpha^2 + r^2 \dot\phi^2 \right.\nonumber\\ & & \phantom{[}\left.\null -2 \dot\phi\dot\alpha r^2 + 2 rr \dot\alpha(\dot\alpha -\dot\phi ) \cos(\phi-\pi ) \right]~,\end{aligned}\ ] ] which contains harmonic , but also strongly nonlinear terms .
if the kinetic energy is expanded around the energy minimum , we see that all non - harmonic terms originate from the @xmath27 factors . for @xmath28 , these higher order terms become fourth order , whereas for @xmath29 , they are third order . for molecules with bond angles not equal to @xmath30 ,
the nonlinear terms are stronger and can be explored at lower energies .
the geometry produces nonlinearity , even in molecules with pair - wise harmonic potentials , and the strength of the nonlinearity depends on the bond angles .
the considerations above suggest that one should look for chaos in molecules with a sufficient number of atoms , and with somewhat complicated geometry .
molecules such as carbon dioxide , which has the right kind of geometry , but only three atoms , may not be sufficiently nonlinear for energies corresponding to room temperature .
larger molecules , for instance benzene , have more degrees of freedom , and are more likely to produce chaos .
consider the atomistic model of benzene shown in figure [ fig : benzene ] .
the angles between bonds are @xmath31 , and the six heavy carbon atoms alone should have a sufficient number of degrees of freedom to produce chaos . for the interaction between the atoms ,
we use the tripos 5.2 force field described in reference @xcite .
the bending and stretching contribute to the potential energy through quadratic terms around their energy minima . additionally , there is also a somewhat weaker contribution due to torsion , which depends on the positions of 4 atoms .
the hydrogen atoms in benzene are light , and the frequencies associated with various stretching and bending modes of the c - h bonds are high compared to frequencies related to interactions between heavier atoms . even at room temperature ,
the quantum states of hydrogen bonds are far apart compared to the typical available energy .
the associated modes are difficult to excite and these bonds remain in the ground state . in other words ,
the hydrogen bonds do not participate in the chaotic dynamics .
if an atomistic model for benzene is to be used to investigate the chaotic dynamics , the c - h bonds must first be eliminated .
the hydrogen atoms can be averaged out using a mean - field approximation , by assuming that a hydrogen atom is , on average , in the same plane as the three nearest carbon atoms , i. e. the one to which it is bonded and its two neighbours .
this greatly simplifies the torsion terms .
let @xmath32 and @xmath33 denote the position and momentum of the @xmath34-th carbon atom , ordered in such a way that @xmath34 and @xmath35 are neighbours .
let @xmath36 be the angle between @xmath37 and @xmath38 , and @xmath39 the torsion angle associated with the bonds between the @xmath40-st , @xmath34-th , @xmath35-st , and @xmath41-nd carbon atoms .
the hamiltonian can be written as @xmath42~,\end{aligned}\ ] ] where @xmath23 and @xmath43 are the effective bending force constant and the effective torsion constant .
the effective bending constant contains contributions from c - c - c , c - c - h , and h - c - c angles .
the effective torsion constant consists of the four torsion terms for the combinations c - c - c - c , c - c - c - h , h - c - c - c , and h - c - c - h , @xmath44 the constants from reference @xcite , which will be used in this model for the rest of this paper are @xmath45 , @xmath46@xmath47 , @xmath48 , and @xmath49 . we have performed molecular dynamics simulations of this model of benzene and numerically obtained the largest lyapunov exponent , which quantifies the chaotic behaviour .
the system was started from random initial conditions and integrated using the velocity - verlet algorithm .
the velocity - verlet algorithm is symplectic and ensures that the conserved quantities in hamiltonian systems remain conserved for finite time - step size @xcite . for a nonlinear dynamical system with a large number of dimensions
it is impossible to determine numerically within reasonable time the set of initial conditions that have a specific energy .
we therefore choose the initial conditions with zero total momentum and zero angular momentum , and on the intersection of this with the constant energy shell of the linearised system , with energy corresponding to a specific temperature .
for all temperatures at which the classical description is valid , the lyapunov exponents were found to be positive . at energy corresponding to room temperature , 32 set of initial conditions were investigated , and all of them produced chaos .
we have extended our atomistic molecular dynamics simulations of the benzene molecule described before by adding an external potential to represent the coupling to the hexagonal lattice of a graphite substrate .
we choose graphite , because of its weak interaction with adsorbed molecules . each carbon atom is affected separately by the potential .
the inter - atomic distance in the graphite was chosen to be equal to 1.42 and the corrugation was 25 mev . in the direction perpendicular to the surface ,
the potential was harmonic with spring constant 61.2 mev / @xmath50 .
the theoretical expressions for the effective diffusion and drift can now , in principle , be evaluated , along the lines of the calculations in @xcite .
though the qualitative behaviour of the system is derived in this paper using very simple arguments , it should be noted that the quantitative evaluation of the theoretical expressions for the diffusion and friction is non - trivial , because the benzene molecule is far more complicated than the bouncing ball of reference @xcite or the hydrogen atom of reference @xcite .
initial conditions were again chosen randomly , with the energy of the linearised system corresponding to room temperature , and with zero total momentum .
the model molecule was placed on a substrate in an energy minimum , and the system was integrated with time steps of size of 1.0 a.u=0.0242 fs .
the motion of the centre of mass is displayed in figure [ fig : trajectories ] for eight different sets of initial conditions .
all eight systems have a positive largest lyapunov exponent .
though the initial conditions for the centre of mass are identical in all cases , the eight resulting trajectories are different . the chaotic internal dynamics therefore cause diffusion of the centre of mass on the substrate .
analysis of the time - dependence of the mean square displacement suggests that the diffusion is normal .
however , the diffusion sometimes exhibits long jumps , which produces a dependence of the mean square displacement on time which only becomes linear for long times and for two trajectories displayed in the figure [ fig : trajectories ] the analysis is inconclusive .
note that in a system which is also subject to thermal noise , extremely long jumps could be more rare .
the internal degrees of freedom of molecules can be related to the diffusion of their centres of mass on a substrate .
if the internal degrees of freedom are chaotic , they can be seen as a finite heat bath which generates noise that leads to diffusion of the centre of mass .
this enhances the diffusion of a molecule with internal degrees of freedom on a substrate compared to diffusion with purely thermal origin .
we have demonstrated in this paper that atomistic models of molecules with pair - wise harmonic potentials can produce chaotic behaviour . in a realistic model of benzene
, chaos occurs at energies comparable to realistic temperatures .
we have shown that the time - scale separation combined with internal chaos of the benzene molecule implies that the benzene molecule should diffuse on a sufficiently weak substrate , such as graphite without any external sources of noise .
the preliminary numerical studies of benzene on graphite presented here confirm this .
these results raise the possibility that the time - scale separation between the internal dynamics and the centre - of - mass motion could enhance the diffusion of heavy molecules in experiments . within the frame - work of the time - scale separation theory ,
theoretical expressions exist for the effective diffusion of such molecules due to their internal degrees of freedom .
however , evaluating these expressions is not a trivial exercise , as even the simple model of the benzene molecule on a graphite surface described in this paper is more complicated than the systems for which they have been evaluated in the past .
additionally , the interplay between the noise from the finite heat bath and the thermal noise from the environment will have to be investigated further .
asw would like to thank nwo ( the netherlands organisation for scientific research ) for financial support through the veni program .
10 m. eichberger , m. marschall , j. reichert , a. weber - bargioni , w. auwrter , r. l. c. wang , h. j. kreuzer , y. pennec , a. schiffrin , and j. v. barth .
dimerization boosts one - dimensional mobility of conformationally adapted porphyrins on a hexagonal surface atomic lattice .
, 8:46084613 , 2008 .
ju , w .- j .
lee , h .- c .
chen , and j .-
molecular dynamics simulation of the dynamical behaviors of an aromatic carboxylic acid molecule with different conformations on a au(111 ) surface .
, 90:143112 , 2007 .
w. c. swope , h. c. andersen , p. h. berens , and k. r. wilson .
a computer simulation method for the calculation of equilibrium constants for the formation of physical clusters of molecules : application to small water clusters .
, 76:637 , 1982 . | chaotic internal degrees of freedom of a molecule can act as noise and affect the diffusion of the molecule on a substrate .
a separation of time scales between the fast internal dynamics and the slow motion of the centre of mass on the substrate makes it possible to directly link chaos to diffusion .
we discuss the conditions under which this is possible , and show that in simple atomistic models with pair - wise harmonic potentials , strong chaos can arise through the geometry . using molecular - dynamics simulations , we demonstrate that a realistic model of benzene is indeed chaotic , and that the internal chaos affects the diffusion on a graphite substrate . |
synchronization of coupled oscillators is important and has been widely studied in variety of systems from physics , chemistry and biology @xcite .
partially locked states are states of imperfect synchronization with locked subpopulation and drifting subpopulation , and have been observed in coupled oscillators with distributed intrinsic frequencies @xcite .
the formation of locked subpopulation with the increase of the coupling strength is one of the explanations for the transition between asynchronous states and synchronous states . in these cases ,
partial locking is due to the inhomogeneity in intrinsic frequencies .
for some oscillators , the coupling strength is strong enough to make the oscillators locked to the coupling force by overcoming the desynchronizing effect of frequency difference , but for others it is not strong enough and they drift . however , there can be desynchronizing factors other than ones intrinsic to uncoupled oscillators .
recent studies of networks of real systems show connection topology and coupling strength can be far from homogeneous @xcite . especially , in many real systems including the world - wide web , the internet , social and biological networks , the number of connections per node or the total coupling strength per node follows scale - free distribution ( or power - law distribution ) @xmath0 @xcite .
theoretical studies have shown that the inhomogeneity in coupling can induce asynchronous states or make synchronization harder to achieve @xcite . in this paper , we study the dynamics of oscillators with coupling strength which has a scale - free distribution .
it is shown that partially locked states can occur due to the cooperation of the inhomogeneity and the coupling function .
in contrast to the previously studied partially locked states in systems with distributed intrinsic frequencies , the partially locked states due to coupling inhomogeneity can be bistable with synchronous states . using a self - consistency argument
, we analytically obtain various values of the partially locked states .
to focus on the coupling inhomogeneity effect , we consider systems of coupled identical limit cycle oscillators with same type of coupling . in the case of weak coupling , the systems can be reduced to the following phase - only ones @xcite .
@xmath1 where @xmath2 is the phase of oscillator @xmath3 at time @xmath4 , @xmath5 is the natural frequency of the oscillators , and @xmath6 is the total number of oscillators .
@xmath7 is the coupling strength from oscillator @xmath8 to oscillator @xmath3 and @xmath9 .
@xmath10 is the coupling function obtained by the phase reduction method for pair - wise interaction @xcite .
recently , we introduced the following mean - field model as an approximation of the model of eq .
( [ eq_model_0 ] ) @xcite .
@xmath11 where @xmath12 corresponds to the average coupling strength to oscillator @xmath3 .
this model is a simple generalization of the kuramoto model @xcite where @xmath13 is the same for all the oscillators , and the coupling inhomogeneity is incorporated in @xmath13 . due to the mean - field coupling ,
this model , like the kuramoto model , is easy to simulate and analyze . in the following sections , we use this model to study the scale - free coupling inhomogeneity and
relate the results with those obtained from the simulations with scale - free networks @xcite . here , we use @xmath14 with @xmath15 and @xmath16 instead of @xmath17 of the kuramoto model @xcite . in most of the previous studies with coupled phase oscillators on scale - free networks , models with @xmath17
have been studied @xcite . but coupling functions of the form @xmath14 are more realistic approximations of those obtained from coupled limit cycle oscillators @xcite , and @xmath18 can affect the dynamics significantly in the systems with inhomogeneity in the number of inputs or in the coupling strength to an oscillators @xcite .
@xmath19 is an approximate coupling function for diffusively coupled oscillators such as gap - junction coupled oscillating neurons with which the coupling is zero when the two oscillators are at the same point in the phase space @xcite . with the coupling function @xmath19 , @xmath16 ,
in - phase synchrony is a solution of eq .
( [ eq_model_1 ] ) regardless of the coupling strengths , and the state is locally stable since @xmath20 ( theorem 3.1 of ref .
@xcite ) .
we perform numerical simulations of eq .
( [ eq_model_1 ] ) with @xmath19 using a fourth order runge - kutta method with time step @xmath21 . unless noted otherwise , we use near uniformly incoherent initial conditions : @xmath22 is chosen randomly from @xmath23 for all @xmath3 .
note that near in - phase synchronous initial conditions lead to the locally stable in - phase synchrony . using the rejection method @xcite , we randomly select @xmath13 according to a truncated scale - free distribution @xmath24 . @xmath25 $ } , \\ 0 , & \mbox{otherwise},\\ \end{array } \right .
\label{eq_dist_k}\end{aligned}\ ] ] where @xmath26 and @xmath27 is a normalization factor that satisfies the normalization condition @xmath28 .
we obtain @xmath29 .
truncation in the distribution is introduced to ensure numerical stability but simulations with larger @xmath30 show that untruncated scale - free distribution gives only quantitatively different results . after the assigning of all the values of @xmath13 ,
we renumber the oscillators according to the ascending order of coupling strength to clearly see the dynamics dependence on @xmath13 . to measure the degree of synchrony
, we calculate @xmath31 @xmath32 is the order parameter showing the degree of synchrony : @xmath32 is between @xmath33 and @xmath34 , with @xmath33 meaning uniform incoherence and @xmath34 in - phase synchrony . in the simulations , @xmath32 approaches a stationary value with small fluctuations and the time average of @xmath32 is calculated .
figure [ fig_phs ] shows simulation results of eq .
( [ eq_model_1 ] ) . in figs . [ fig_phs](a ) and ( b ) , coupling strength @xmath13 and the distribution of @xmath13 are plotted , respectively . @xmath35 and @xmath36
are used in the simulations .
figures [ fig_phs](c ) and ( d ) are the snapshot of the phases ( @xmath37 ) and the frequency ( @xmath38 ) of oscillators , respectively .
while in - phase synchrony is reached from almost all initial conditions for the cases with uniform coupling ( @xmath39 ) @xcite , the system evolves to a partially locked state with the coupling inhomogeneity for some range of @xmath40 value ( figs . [ fig_phs](c ) and ( d ) ) .
the oscillators are divided into a phase - locked group giving a continuous line in the figures and drifting group giving scattered dots . because the oscillator indices are renumbered according to the order of the coupling strength , the locked oscillators and the drifting ones are clearly distinguished . with the coupling strength distribution treated in this study ,
it is observed that oscillators with small @xmath13 are locked ( figs .
[ fig_phs](a ) , ( c ) , and ( d ) ) .
this is a new phenomenon . while most of partially locked states discussed previously were due to the inhomogeneity in the natural frequencies of oscillators @xcite , those of this study are mainly due to the coupling inhomogeneity . without the inhomogeneity ,
the states can not exist .
the formation of partially locked states in this system may be related to the formation of so - called chimera states in nonlocally coupled identical oscillators with a similar coupling function @xcite . in chimera states which can be classified as partially locked states , phase - locked oscillators spatially
coexist with drifting ones . in those systems , there is no coupling inhomogeneity and thus @xmath18 of the coupling function has no significant role . in both of the cases ,
partially locked states occur when @xmath40 is near @xmath41 @xcite .
we can see more of the similarity in the next section through analysis and simulation .
note also that the partially locked states are bistable with in - phase synchronous states .
the bistability between a chimera state and an in - phase synchronous state is a similar one @xcite . for @xmath42 ,
in - phase synchrony is not a solution of the system and unless @xmath18 is large enough to induce uniform incoherence , we can obtain partially locked states with the truncated scale - free distribution or with other distributions @xcite .
in this section , we analyze the partially locked states using a self - consistency argument @xcite . before analyzing partially locked states ,
we need to look into the local stability of uniformly incoherent states , because uniformly incoherent states are solutions of the model and these states can compete with partially locked states . in recent paper @xcite
, using population density analysis , we showed that for the model of eq .
( [ eq_model_1 ] ) with @xmath43 , the eigenvalues @xmath44 determining the stability of an incoherent state satisfy the equations @xmath45 where @xmath24 is the distribution for @xmath46 . for the coupling strength distribution of eq .
( [ eq_dist_k ] ) , these equations become @xmath47 in the limit of @xmath48 , we can find critical @xmath18 above which uniformly incoherent state is obtained @xcite .
if we take the conditions @xmath49 and @xmath50 without which eqs .
( [ eq_lambda_r1 ] ) and ( [ eq_lambda_i1 ] ) can not be satisfied , the equations become the following in the limit . @xmath51 figure [ fig_c_cr ] shows critical @xmath18 ( @xmath52 ) as a function of @xmath40 .
the critical values denoted by solid lines are obtained numerically from eqs .
( [ eq_cos_final ] ) and ( [ eq_sin_final ] ) .
when @xmath53 , incoherent states are unstable . for the case of eq .
( [ eq_model_1 ] ) with @xmath15 and @xmath54 ) studied in this paper , @xmath18 is less than @xmath52 as shown in fig .
[ fig_c_cr ] and thus the incoherent states are unstable .
smaller @xmath55 gives ranges of @xmath40 values with @xmath56 , but for the simplicity of the discussion we restrict the cases to those with @xmath57 . since incoherent states are unstable in this situation , we get states other than incoherent states even when initial conditions are near - uniformly incoherent states .
the system evolves to a partially locked state or an in - phase synchronous state from near - uniformly incoherent states .
now , let us analyze the partially locked states .
kuramoto used self - consistency arguments and derived an equation for the order parameter @xmath32 to analyze his model in which oscillators with distributed frequencies are globally coupled @xcite .
the idea is to calculate the order parameter by calculating the contribution from locked subpopulation and from drifting subpopulation in stationary states .
because the calculation of the contribution from the subpopulations contains the order parameter , this gives a self - consistent equation for the order parameter .
a similar argument with space - dependent order parameter was used by kuramoto and his colleagues to analyze recently observed chimera states in nonlocally coupled identical oscillators @xcite . in chimera states ,
phase - locked oscillators coexist with drifting ones @xcite .
we use the same self - consistency argument here to analyze our system . in this case
@xmath32 is the same for all the oscillators as in the kuramoto model , but the coupling term is different for each oscillator as in chimera states because of the dependency of the term on the coupling strength @xmath13 .
this inhomogeneity in the coupling terms causes the splitting of the population into two groups - locked group and drifting group .
we follow the same argument for the analysis of chimera states described in detail in ref .
the main difference is that coupling strength distribution @xmath24 is used instead of coupling kernel and the order parameter @xmath32 is the same for all oscillators in our case .
we assume the limit of infinitely many oscillators and focus only on stationary states .
let @xmath58 denote the frequency of the population oscillation of eq .
( [ eq_op ] ) after the system approaches a stationary state and @xmath59 represent the phase of oscillators relative to the average oscillation
. then we can rewrite eq .
( [ eq_model_1 ] ) using the order parameter @xmath60 defined in eq .
( [ eq_op ] ) as follows .
@xmath61 , \nonumber \\ & & ~i = 1,2 , ... , n , \label{eq_sine_1}\end{aligned}\ ] ] where @xmath62 .
when the system reaches a stationary state , @xmath32 and @xmath63 do not depend on time .
the oscillators with @xmath64 asymptotically approach a stable fixed point @xmath65 obtained from the following equations .
@xmath66 and a stability condition for the fixed point @xmath67 these oscillators are those which are phase - locked at frequency @xmath58 in the original frame . the oscillators with @xmath68 drift monotonically .
this subpopulation can be described by an invariant probability density @xmath69 in the stationary state .
the invariant probability density should satisfy the condition @xmath70 , where @xmath71 is the instantaneous frequency @xmath72 .
therefore , we get @xmath73 where the normalization constant is chosen such that @xmath74 . in the rotating frame of the population oscillation with the frequency @xmath58 ,
the order parameter contribution from locked subpopulation is calculated as follows .
@xmath75 where @xmath76 and @xmath77 is the domain with @xmath78 .
we use eqs .
( [ eq_locked ] ) and ( [ eq_locked_stabcond ] ) to calculate @xmath79 .
the order parameter contribution from drifting subpopulation can be calculated by using population density @xmath69 of eq .
( [ eq_density ] ) .
@xmath80 , \nonumber \\ \label{eq_op_drift}\end{aligned}\ ] ] where @xmath81 is calculated using contour integration , @xmath82 is the sign function and @xmath83 .
@xmath84 appears here , because it determines which pole lies inside the contour .
@xmath85 is the domain with @xmath86 .
the order parameter @xmath87 in the rotating frame is the sum of the contributions from locked subpopulation ( eq .
( [ eq_op_lock ] ) ) and from drifting subpopulation ( eq . ( [ eq_op_drift ] ) ) .
because @xmath32 and @xmath63 are independent of @xmath46 , we obtain @xmath88 , \nonumber \\
\label{eq_r2}\end{aligned}\ ] ] where @xmath89 is the total range for @xmath46 .
this gives two independent equations for @xmath32 and @xmath90 .
@xmath91 corresponding to the in - phase synchronous state and @xmath92 corresponding to the uniformly incoherent state are solutions of eq .
( [ eq_r2 ] ) .
but note that this fact does not guarantee the stability of each state .
we numerically find the solutions @xmath93 which correspond to the partially locked states of the system . in the cases with coupling strength distribution of eq .
( [ eq_dist_k ] ) , numerical simulations of the model show that @xmath46 with which oscillators are locked are bounded above , @xmath90 is negative , and ( @xmath94 ) is negative .
based on the simulations and the condition for the locking , we can guess that @xmath95 and @xmath96 .
in addition , @xmath97 is positive in the domain @xmath85 . using the observation from simulations ,
we numerically obtain @xmath32 and @xmath90 from eq .
( [ eq_r2 ] ) .
we also compute the fraction of drifting oscillators . @xmath98 in figs .
[ fig_analysis](a ) , ( b ) , and ( c ) , we plot the values of @xmath32 , @xmath90 , and @xmath99 obtained both from simulations and analysis for partially locked states .
the analysis ( solid lines ) shows good agreement with simulation results ( symbols ) .
as mentioned , the in - phase state is stable for the entire parameter range of this system ( not shown in the figures ) .
we use near - incoherent states as initial conditions in simulations and get partially locked states for @xmath100 which depends on @xmath55 . for @xmath101 ,
the system reaches to in - phase synchronous state .
simulations and analysis show that @xmath102 becomes larger as @xmath55 increases .
this is consistent with the fact that as @xmath55 increases , the coupling distribution becomes that of the case with uniform coupling strength and in - phase synchronous states are asymptotically reached from almost all initial conditions in the case with uniform coupling strength @xcite .
the partially locked states exist for @xmath40 values near @xmath41 , which is similar parameter range for the existence of chimera states in the systems with nonlocal coupling @xcite . as for the chimera states , the transition points to partially locked states appear abruptly .
this contrasts to the continuous transition between partially locked states and asynchronous states in systems with distributed frequencies @xcite .
note that this system has bistability between an in - phase synchronous state and an partially locked state . to see how the selection of the state between in - phase state and partially locked state depends on the initial conditions in the presence of inhomogeneity @xcite
, we look at the asymptotic state as a function of the initial conditions .
the initial values for @xmath103 are randomly chosen from @xmath104 , where @xmath105 $ ] .
the corresponding initial value of @xmath32 is given by @xmath106 .
we scan the interval @xmath107 $ ] for @xmath108 with step size @xmath109 to find @xmath108 below which the system goes to the in - phase synchronous state with @xmath109 intervals .
we denote the @xmath108 value @xmath110 .
we average the corresponding @xmath111 values ( @xmath112 values ) over states from 10 different configurations . in fig .
[ fig_analysis](d ) , the filled symbols show the @xmath112 values . in the simulations , we obtain partially locked states for @xmath113 and in - phase synchronous states for @xmath114 .
no other states are obtained in our simulations .
open symbols and solid curves are the same ones as in fig .
[ fig_analysis](a ) .
each dashed curve represents the other nontrivial branch of the solution of eq .
( [ eq_r2 ] ) .
the values of @xmath112 are close to but below the dashed curves .
the states for the dashed curves seem to be unstable and seem to act as basin boundaries for at least this type of initial conditions .
the fact that the @xmath112 values are close to @xmath34 shows that the basin of attraction for the partially locked states is relatively large compared to that of the in - phase synchronous states and thus the partially locked states can occur rather naturally in this system .
synchronous states are stable even with the inhomogeneity but the inhomogeneity makes the size of the sync basin smaller @xcite .
the results of previous sections are applicable in systems of coupled oscillators on networks .
@xmath115 \nonumber \\ & & ~i = 1,2 , ... , n , \label{eq_sf_model}\end{aligned}\ ] ] where oscillator @xmath3 is influenced by @xmath116 neighbors with coupling strength @xmath46 according to a coupling topology described by an adjacency matrix @xmath117 .
@xmath116 is called the degree of @xmath3 .
we take the element of adjacency matrix @xmath118 , if oscillator @xmath8 influences oscillator @xmath3 , and @xmath119 otherwise .
we consider the cases of truncated scale - free networks @xcite where the degree distribution of @xmath116 , @xmath120 , is given by @xmath121$},\\ 0 , & \mbox{otherwise}.\\ \end{array } \right .
\label{eq_ki}\end{aligned}\ ] ] when oscillators are randomly coupled to others and @xmath122 is sufficiently large , we can use the following approximation for eq .
( [ eq_sf_model ] ) .
@xmath123 with this , eq . ( [ eq_sf_model ] ) is approximately equivalent to @xmath124 , \label{eq_sf_approx_model}\end{aligned}\ ] ] which is eq .
( [ eq_model_1 ] ) with @xmath13 being @xmath125 .
we simulate eq .
( [ eq_sf_model ] ) with networks following a given degree distribution @xmath120 of eq .
( [ eq_ki ] ) .
the networks are generated as follows . using the similar method of section
ii , we randomly select a positive integer @xmath126 $ ] and assign it to an oscillator among the @xmath6 oscillators as the degree of the oscillator .
we randomly select @xmath127 oscillators as the neighbors .
after the network is generated , the oscillators are renumbered according to the ascending order of the degree . for simplicity of generating networks ,
we use directed networks but bidirectional networks do not change the results significantly . figure [ fig_sf_phs ] shows the simulation results of eq .
( [ eq_sf_model ] ) .
figure [ fig_sf_phs](a ) shows the degree distribution of the network .
figures [ fig_sf_phs](b ) and ( c ) are the snapshots of the phases and the frequency of oscillators , respectively . as in the cases with inhomogeneous coupling strength shown previous sections , this system also shows similar states . in this system ,
the states have near locked oscillators ( figs .
[ fig_sf_phs](b ) and ( c ) ) .
we compare the simulation results of this system with those obtained from eq .
( [ eq_r2 ] ) using the fact @xmath13 corresponds to @xmath125 .
figure ( [ fig_sf_phs])(d ) shows good agreement between them .
in summary , we have investigated coupled identical oscillators with scale - free distribution of coupling strength and found that partially locked states can occur due to the inhomogeneity and the coupling function .
various quantities of the partially locked states have been computed through a self - consistency argument .
this study contrasts with the previous studies in the fact that the coupling inhomogeneity instead of the frequency inhomogeneity is the main cause of partial locking and partially locked states can be bistable with synchronous states .
our findings may help further understanding of synchronous behavior on inhomogeneous networks .
this work was supported by national science foundation grant dms05135 .
et al . _ ,
plos biology * 3 * , 0507 ( 2005 ) .
d. golomb and d. hansel , neural comp .
* 12 * , 1095 ( 2000 ) .
m. denker , m. timme , m. diesmann , f. wolf , and t. geisel , phys .
lett . * 92 * , 074103 ( 2004 ) .
t. nishikawa , a. e. motter , y .- c .
lai , f. c hoppensteadt , phys .
rev . lett . * 91 * , 014101 ( 2003 ) .
a. e. motter , c. zhou , and j. kurth , europhys . lett . *
69 * , 334 ( 2005 ) ; phys .
e * 71 * , 016116 ( 2005 ) . c. zhou , a. e. motter , and j. kurths , phys .
lett . * 96 * , 034101 ( 2006 ) .
ko and g. b. ermentrout ( submitted ) .
j. g. restrepo , e. ott , and b. r. hunt , phys .
e * 71 * 036151 ( 2005 ) ; chaos * 16 * , 015107 ( 2005 ) ; phys . rev
. lett . * 96 * 254103 ( 2006 ) .
p. n. mcgraw and m. menzinger , phys .
e * 72 * , 015101(r ) ( 2005 ) .
s. watanabe and s. h. strogatz , phys .
. lett . * 70 * , 2391 ( 1993 ) . s. watanabe and s. h. strogatz , physica d * 74 * , 197 ( 1994 ) .
y. kuramoto and d. battogtokh , nonlin .
compl . syst . * 5 * , 380 ( 2002 ) .
d. m. abrams and s. h. strogatz , phys .
lett . * 93 * , 174102 ( 2004 ) .
d. m. abrams and s. h. strogatz , int .
j. bif . and chaos * 16 * , 21 ( 2006 ) .
d. a. wiley , s. h. strogatz , and m. girvan , chaos * 16 * , 015103 ( 2006 ) . | we investigate coupled identical phase oscillators with scale - free distribution of coupling strength . it is shown that partially locked states can occur due to the inhomogeneity in coupling and some properties of the coupling function .
various quantities of the partially locked states are computed through a self - consistency argument and the values show good agreement with simulation results . |
why are new calculations of standard solar models of interest ? after all , solar models have been used to calculate neutrino fluxes since 1962 ( bahcall , fowler , iben , & sears 1963 ) and solar atmospheres have been used to calculate p - mode oscillation frequencies since 1970 ( ulrich 1970 ; leibacher & stein 1971 ) . over the past four decades , the accuracy with which solar models are calculated has been steadily refined as the result of increased observational and experimental information about the input parameters ( such as nuclear reaction rates and the surface of abundances of different elements ) , more accurate calculations of constituent quantities ( such as radiative opacity and equation of state ) , the inclusion of new physical effects ( such as element diffusion ) , and the development of faster computers and more precise stellar evolution codes .
solar models nevertheless remain at the frontiers of two different scientific disciplines , solar neutrino studies and helioseismology . in an era in which many major laboratory studies are underway to study neutrino oscillations with the aid of very long baselines , @xmath1 km , between accelerator and detector ,
solar neutrinos have a natural advantage , with a baseline of @xmath2 km ( pontecorvo 1968 ) .
in addition , solar neutrinos provide unique opportunities for studying the effects of matter upon neutrino propagation , the so - called msw effect ( wolfenstein 1978 ; mikheyev & smirnov 1985 ) , since on their way to terrestrial detectors they pass through large amounts of matter in the sun and , at night , also in the earth .
the connection with ongoing solar neutrino research imposes special requirements on authors carrying out the most detailed solar modeling .
precision comparisons between neutrino measurements and solar predictions are used by many physicists to refine the determination of neutrino parameters and to test different models of neutrino propagation .
since the neutrino experiments and the associated analysis of solar neutrino data are refined at frequent intervals , it is appropriate to reevaluate and refine the solar model predictions as improvements are made in the model input parameters , calculational techniques , and descriptions of the microscopic and macroscopic physics . in this paper
, we provide new information about the total solar neutrino fluxes and the predicted neutrino event rates for a set of standard and non - standard solar models . using the best - available standard solar model , we also present the calculated radial dependence of the production rate for each of the important solar neutrino fluxes .
we publish for the first time the results of a precision calculation with the standard solar model of the electron density throughout the sun , from the innermost regions of the solar core to the solar atmosphere .
we also present for the first time a detailed calculation of the radial profile of the number density of scatterers of sterile neutrinos .
these quantities are important for precision studies of neutrino oscillations using solar neutrinos .
we also provide detailed predictions for the time evolution of some of the important solar characteristics such as the depth and mass of the solar convective zone , the radius and the luminosity of the sun , the central temperature , density , pressure , and hydrogen mass fraction , as well as the temperature , density , pressure , and radiative opacity at the base of the convective zone .
as far as we know , these are the first detailed results submitted for publication on the time evolution of many of these quantities
. some of the calculated time dependences may be subject to observational tests . at the present writing ,
the sun remains the only main - sequence star for which p - mode oscillations have been robustly detected .
thus only for the sun can one measure precisely tens of thousands of the eigenfrequencies for stellar pressure oscillations .
the comparison between the sound speeds and pressures derived from the observed p - mode frequencies and those calculated with standard solar models has provided a host of accurate measurements of the interior of the nearest star .
the solar quantities determined by helioseismology include the sound velocity and density as a function of solar radius , the depth of the convective zone , the interior rotation rate , and the surface helium abundance .
the excellent agreement between the helioseismological observations and the solar model calculations has shown that the large discrepancies between solar neutrino measurements and solar model calculations can not be due to errors in the solar models ( cf .
figure [ fig : diffbp00best ] ) . in this paper , we present a refined comparison between our best standard solar model and measurements of the solar sound speeds obtained using oscillation data from a number of different sources .
we use a combination of data from the lowl instrument and the bison network , two sets of data from the golf instrument , as well as data from the gong network and the mdi instrument .
we describe in the following paragraph the organization of this paper .
but , since this paper contains a lot of information on disparate topics , we recommend that the reader first turn to the final section , [ sec : discussion ] , and peruse our summary and discussion of the main new results .
section [ sec : discussion ] may help the reader decide which sections of the paper the reader wants to read ( or to skip ) .
the different sections are written so that they can be read more or less independently .
section [ sec : standard ] defines the slightly - improved standard solar model and describes the numerical tables that present details of the contemporary characteristics of the standard model .
section [ sec : timedependences ] describes the main sequence time dependences of some of the most important characteristics of the standard model .
we present in this section the time evolution of the solar radius and luminosity , the properties of the convective zone , and the physical characteristics of the center of the sun .
we discuss solar properties from the initial age main sequence to an age of 8 billion years .
section [ sec : variantdeviant ] defines and compares the physical characteristics of seven variant standard models and four deviant ( deficient ) solar models , which together with the standard model make up a set of twelve models whose neutrino fluxes we evaluate in this paper .
we previously used a subset of nine of these models to test the robustness of helioseismological inversions ( basu , pinsonneault , & bahcall 2000 , hereafter bpb2000 ) .
we discuss in this same section two new standard - like solar models with heavy element to hydrogen ratios that differ slightly from our previously adopted value of @xmath3 .
section [ sec : neutrinos ] discusses solar neutrino physics .
we present the predicted present - day neutrino fluxes for the standard model and for all of the variant and deviant solar models , as well as the electron number density versus solar radius .
we contrast the predicted neutrino event rates with the results of the chlorine , kamiokande , gallex , sage , super - kamiokande , and gno solar neutrino experiments .
we also give in this section the calculated time evolution of the most important solar neutrino fluxes .
section [ sec : soundspeeds ] compares calculated and observed sound speeds .
we present the results both on a panoramic scale suitable for discussing the implications for solar neutrino physics and on a zoomed - in scale appropriate for detailed investigations of solar physics . in [ sec : hepluscz ] , we compare , for all 12 of the solar models discussed in this paper , the calculated values of the surface helium abundance and the depth of the convective zone with the measured values for these quantities .
we summarize and discuss our main results in [ sec : discussion ] . the interested reader may wish to consult the following works that summarize the solar neutrino aspects of solar models ( bahcall 1989 ; bahcall & pinsonneault 1992 , 1995 ; berezinsky , fiorentini , & lissia 1996 ; castellani et al .
1997 ; richards et al . 1996 ; turck - chize et al .
1993 ; bahcall , basu , & pinsonneault 1998 ) and the helioseismologic aspects of solar models ( bahcall & ulrich 1988 ; bahcall & pinsonneault 1995 ; christensen - dalsgaard et al .
1996 ; ; guenther & demarque 1997 ; guzik 1998 ; turck - chize et al .
1998 ; brun , turck - chize , & zahn 1999 ; fiorentini & ricci 2000 ) .
we define in [ subsec : definitionstandard ] the standard solar model and present in [ subsec : standardcharacteristics ] some of the important contemporary characteristics of the standard solar model , including detailed tables of the physical variables as a function of the solar radius . by ` the standard solar model ' ,
we mean the solar model which is constructed with the best - available physics and input data .
all of the solar models we consider , standard or ` deviant ' models , ( see below ) are required to fit the observed luminosity and radius of the sun at the present epoch , as well as the observed heavy element to hydrogen ratio at the surface of the sun .
no helioseismological constraints are used in defining the standard solar model .
naturally , standard models improve with time , as the input data are made more accurate , the calculational techniques become faster and more precise , and the physical description is more detailed .
thus it is necessary at each presentation of a new standard model to describe what has changed from the previous best model and to provide references to the earlier work .
the reader can see bahcall ( 1989 ) for a general reference to the early work on precise solar models that were motivated by the need to predict accurate solar neutrino fluxes and to make comparisons with helioseismological data .
our standard solar model is constructed with the opal equation of state ( rogers , swenson , & iglesias 1996 ) and opal opacities ( iglesias & rogers 1996 ) , which are supplemented by the low temperature opacities of alexander & ferguson ( 1994 ) .
the model was calculated using the usual mixing length formalism to determine the convective flux .
the principal change in the input data is the use of the grevesse & sauval ( 1998 ) improved standard solar composition in the opal opacities ( see http://www-phys.llnl.gov/research/opal/index.htm ) and in the calculation of the nuclear reaction rates .
the refinements in this composition redetermination come from many different sources , including the use of more accurate atomic transition probabilities in interpreting solar spectra .
the opal equation of state and the alexander and ferguson opacities are not yet available with the composition recommended by grevesse and sauval 1998 .
we have used a present - epoch solar luminosity of @xmath4 $ ] for all the models described in detail in this paper . only for the neutrino fluxes presented in [ subsubsec : standardneutrino ] and
[ subsubsec : nacre ] have we used the more recent best - estimate value of solar luminosity , @xmath5 ~=~ 3.842 \times 10^{33 } { \rm ~erg s^{-1}}$ ] ( see frhlich & lean 1998 ; crommelynck , fichot , domingo , & lee 1996 ) . the difference between these two values for the luminosity is @xmath6% . for the calculations of uncertainties in neutrino flux predictions , we assume a @xmath7 uncertainty of @xmath8% .
the uncertainty in the predicted solar neutrino fluxes due to the luminosity is an order of magnitude smaller than the total uncertainty in predicting the neutrino fluxes .
for all the other quantities we calculate , the uncertainty in the solar luminosity has an even smaller effect .
the nuclear reaction rates were evaluated with the subroutine exportenergy.f ( cf .
bahcall & pinsonneault 1992 ) , using the reaction data in adelberger et al .
( 1998 ) and with electron and ion weak screening as indicated by recent calculations of gruzinov & bahcall ( 1998 ) ; see also salpeter ( 1954 ) .
this assumption seriously underestimates the charge density near heavy ions .
for example , it is known that a screened beryllium nucleus under solar interior conditions has charge density near the nucleus @xmath9 ( gruzinov & bahcall 1997 ; brown & sawyer 1997 ; all quantum mechanical calculations give similar results , see bahcall 1962 , and iben , kalata , & schwartz 1967 ) . ] .
the model incorporates helium and heavy element diffusion using the exportable diffusion subroutine of thoul ( cf .
thoul , bahcall & loeb , 1994 ; bahcall & pinsonneault 1995)jnb , menu item : neutrino software and data . ]
. an independent and detailed treatment of diffusion by turcotte et al .
( 1998b ) yields results for the impact of diffusion on the computed solar quantities that are very similar to those obtained here .
we have used the most recent and detailed calculation ( marcucci et al .
2000a ) for the @xmath10-factor for the @xmath11he(p,@xmath12 + @xmath13)@xmath14he reaction : @xmath15 kev b , which is a factor of @xmath16 times larger than the previous best - estimate [ see [ subsubsec : uncertainties ] , bahcall & krastev ( 1998 ) , and marcucci et al .
( 2000b ) for a discussion of the large uncertainties in calculating @xmath17 . for values of @xmath18 in the range of current estimates ,
the assumed rate of the @xmath19 reaction only affects in a noticeable way the calculated flux of @xmath19 neutrinos and does not affect the calculated fluxes of other neutrinos , the helioseismological characteristics , or other physical properties of the sun . for the standard model ,
the evolutionary calculations were started at the main - sequence stage .
the model has a radius of 695.98 mm .
we do not study the pre - main sequence evolution in our standard model .
this epoch has been discussed in the classical work of iben ( 1965 ) and the effects of pre - main sequence have been shown by bahcall & glasner ( 1994 ) to be unimportant for solar neutrino calculations ( see also the thorough discussion by morel , provost , & berthomieu 2000 ) .
we do consider one pre - main sequence model , which differs very little from the corresponding model started at the zero - age main sequence .
the ratio of heavy elements to hydrogen ( @xmath3 ) at the surface of the model is 0.0230 , which was chosen to be consistent with the value obtained by grevesse & sauval ( 1998 ) .
a krishna - swamy @xmath20-@xmath21 relationship for the atmosphere was used .
we adopt a solar luminosity @xmath22 and a solar age of @xmath23 yr ( see bahcall & pinsonneault 1995 ) .
in the course of our recent analysis of systematic uncertainties in helioseismological inversions ( basu , pinsonneault , & bahcall 2000 ) , we uncovered an error in the code we wrote for the interpolation of the opal 95 opacities .
the edges of opacity tables were flagged by zero opacity values ; unfortunately , there were some interpolation problems associated with the occurrence of zero values inside the table .
this problem occurred because the logarithm of the opacity , which is what we were tabulating , can actually be zero .
the interpolation error caused small changes in the opacity which produced errors in the sound speed of order @xmath24% for solar radii in the range of @xmath25 to @xmath26 and errors of order @xmath27% in the @xmath28b neutrino flux ( with smaller errors for other neutrino fluxes ) .
in the following sections , we will point out more specifically the changes in the neutrino predictions and in the sound velocities that are produced by correcting this interpolation error .
these changes are particularly apparent in the comparison of figure [ fig : differences6 ] and figure [ fig : differences5 ] , which are discussed in the helioseismology section ,
[ sec : soundspeeds ] .
the details of the structure of a standard solar model are of importance for both helioseismology and solar neutrino calculations . in the first paper in this series in which we published the details of a model of the solar interior structure ( bahcall et al .
1982 ) , we presented a table with only @xmath29 radial shells ( rows of numbers ) and @xmath30 variables for each shell ( mass , radius , temperature , density , hydrogen fraction , helium fraction , luminosity generated , and the source density of the @xmath31 , @xmath32be , and @xmath28b neutrinos ) . over the years , much greater precision was required by the increasing sophistication of the helioseismological measurements and the solar neutrino calculations .
fortunately , the computing capacity more than made up for the necessary increase in model details . we have created from the output of the present calculations two exportable computer files that contain in easily readable form the details of our best standard solar model ( bp2000 ) .
physical variables are given at @xmath33 separate radial shells , which is the same number of shells used to calculate the solar interior model .
in addition to the variables cited above , this file contains the pressure , electron number density , the mass fractions of @xmath11he , @xmath32be , @xmath34c , @xmath35n , and @xmath36o , as well as the source densities of all eight of the most important solar neutrino fluxes .
these files are accessible at http://www.sns.ias.edu/@xmath37jnb .
previous standard solar models in this series ( published in 1982 , 1988 , 1992 , 1995 , as well as 1998 ) are available at the same url and can be used to test the robustness of conclusions that depend upon models of the solar interior .
in this section , we present and discuss the time dependences of some of the principal characteristics of the standard solar model .
we begin in [ subsec : radiusetal ] by describing the separate temporal evolution of the solar radius and the solar luminosity and then discuss the simple relation we have found between the values of these two quantities as a function of time .
we also demonstrate that the solar luminosity is a robust function of time . in [ subsec : energyfractions ] , we present the time - dependent fractions of the solar luminosity that are produced by different nuclear fusion reactions .
we concentrate in [ subsec : convective ] on the convective zone , presenting the calculated time dependences of the depth and the mass of the convective zone .
we also report the time dependence of the temperature , density , pressure , and opacity at the base of the convective zone . in [ subsec : centralvalues ] , we discuss the time dependence of quantities at the center of the sun , the central temperature , density , pressure , and hydrogen mass fraction . in
[ subsec : separations ] , we calculate and discuss the large and small separations of the @xmath0-mode frequencies as a function of solar age . since we do not discuss pre - main sequence evolution , our calculations are not precise for times less than @xmath38 yr ( see morel , provost , & berthomieu 2000 ; weiss & schlattl 1998 ) .
iben ( 1967,1974 ) and demarque & guenther ( 1991 ) summarize in comprehensive presentations the evolution of solar parameters in models that were calculated prior to the inclusion of element diffusion in solar evolutionary codes .
these discussions did not encounter the problem of semi - convection discussed here in
[ subsec : convective ] because this phenomenon is caused by effects of diffusion near the base of the convective zone .
the solar radius and luminosity ( or equivalently , the solar effective temperature and luminosity ) constitute precise constraints on the possible geological histories of the earth .
we quantify these constraints in the following subsection and specify upper limits to the allowed discrepancies from the standard solar model profile of solar luminosity versus age .
ccccc age&@xmath39&&age&@xmath39 + ( @xmath40 yr)&(@xmath41 ( today))&&(@xmath40 yr)&(@xmath41 ( today ) ) + 0.0 & 0.869 & & 4.2 & 0.987 + 0.2 & 0.882 & & 4.4 & 0.994 + 0.4 & 0.888 & & 4.6 & 1.001 + 0.6 & 0.892 & & 4.8 & 1.008 + 0.8 & 0.897 & & 5.0 & 1.016 + 1.0 & 0.901 & & 5.2 & 1.023 + 1.2 & 0.906 & & 5.4 & 1.031 + 1.4 & 0.910 & & 5.6 & 1.040 + 1.6 & 0.915 & & 5.8 & 1.048 + 1.8 & 0.920 & & 6.0 & 1.057 + 2.0 & 0.924 & & 6.2 & 1.066 + 2.2 & 0.929 & & 6.4 & 1.075 + 2.4 & 0.934 & & 6.6 & 1.085 + 2.6 & 0.940 & & 6.8 & 1.095 + 2.8 & 0.945 & & 7.0 & 1.105 + 3.0 & 0.951 & & 7.2 & 1.116 + 3.2 & 0.956 & & 7.4 & 1.127 + 3.4 & 0.962 & & 7.6 & 1.139 + 3.6 & 0.968 & & 7.8 & 1.152 + 3.8 & 0.974 & & 8.0 & 1.166 + 4.0 & 0.981 + figure [ fig : rvsageplus ] and table [ tab : radiusvstime ] present the calculated radius of the sun versus solar age for the standard solar model .
the values given are normalized to the present - day solar radius , @xmath41 . over the lifetime of the sun ,
the model radius has increased monotonically from an initial value of @xmath42 to the current value @xmath43 , a @xmath44% rise . at a solar age of @xmath45 billion years
, the solar radius will increase to @xmath46 .
we shall see in the following discussion that some of the important evolutionary characteristics of the sun can be expressed simply in terms of the ratio @xmath47 .
ccccc age&@xmath48&&age&@xmath48 + ( @xmath40 yr)&(@xmath49 ( today))&&(@xmath40 yr)&(@xmath49 ( today ) ) + 0.0 & 0.677&&4.2 & 0.970 + 0.2 & 0.721&&4.4 & 0.986 + 0.4 & 0.733&&4.6 & 1.003 + 0.6 & 0.744&&4.8 & 1.020 + 0.8 & 0.754&&5.0 & 1.037 + 1.0 & 0.764&&5.2 & 1.055 + 1.2 & 0.775&&5.4 & 1.073 + 1.4 & 0.786&&5.6 & 1.092 + 1.6 & 0.797&&5.8 & 1.112 + 1.8 & 0.808&&6.0 & 1.132 + 2.0 & 0.820&&6.2 & 1.152 + 2.2 & 0.831&&6.4 & 1.172 + 2.4 & 0.844&&6.6 & 1.193 + 2.6 & 0.856&&6.8 & 1.214 + 2.8 & 0.869&&7.0 & 1.235 + 3.0 & 0.882&&7.2 & 1.256 + 3.2 & 0.896&&7.4 & 1.278 + 3.4 & 0.910&&7.6 & 1.304 + 3.6 & 0.924&&7.8 & 1.332 + 3.8 & 0.939&&8.0 & 1.363 + 4.0 & 0.954 + figure [ fig : lumnormalized ] and table [ tab : lumstandard ] show the calculated solar luminosity as a function of solar age , normalized to the present - day solar luminosity , @xmath49 .
for the standard model , the total solar luminosity has risen monotonically from a zero - age value of @xmath50 .
the time evolution of the solar luminosity is robust .
we also show in figure [ fig : lumnormalized ] the solar luminosity as a function of time for the three most deficient solar models that are described in the following section , [ sec : variantdeviant ] .
the rms difference between the standard luminosity and the luminosity of the deviant models is @xmath51% for the mixed model ( @xmath52% ignoring the first gyr ) , @xmath53% for the no diffusion model ( @xmath54% ignoring the first gyr ) , and @xmath55% for the @xmath56 model ( @xmath57% ignoring the first gyr ) .
the largest deviations occur for the zero - age main sequence models and are @xmath58% for the mixed model , @xmath59% for the no diffusion model , and @xmath60% for the @xmath56 model .
all of the solar models show essentially the same shape for the luminosity evolution as a function of age .
figure [ fig : lumnormalized ] shows that the product @xmath61 to an accuracy of about @xmath62% over the entire period from the zero age main - sequence to an age of @xmath45 billion years .
the solar luminosity @xmath63 varies from @xmath64 to @xmath49 at the present epoch and will reach @xmath65 after @xmath45 billion years on the main sequence .
the corresponding values of @xmath66^{2.5}$ ] are @xmath67 , @xmath68 , and @xmath69 .
thus the luminosity of the sun varies by slightly more than a factor of two while @xmath66^{2.5}$ ] varies by only a few percent .
over the first 5 billion years , eq . ( [ eqn : lrproportional ] ) is satisfied to an accuracy of @xmath70 . from 4 billion years to 8 billion years , the relation is somewhat less steep , @xmath71 to an accuracy of @xmath72 .
this transition from @xmath73 to @xmath74 can be seen clearly in figure [ fig : lumnormalized ] .
lcccc & @xmath75&&&@xmath76 + 0.0 & 0.973 & & 4.5 & 1.000 + 0.5 & 0.983 & & 5.0 & 1.001 + 1.0 & 0.985 & & 5.5 & 1.002 + 1.5 & 0.987 & & 6.0 & 1.003 + 2.0 & 0.990 & & 6.5 & 1.003 + 2.5 & 0.992 & & 7.0 & 1.003 + 3.0 & 0.994 & & 7.5 & 1.001 + 3.5 & 0.996 & & 8.0 & 1.000 + 4.0 & 0.998 + table [ tab : teff ] gives the calculated effective temperature of the standard model , @xmath77 , as a function of solar age .
the effective temperature varies by only @xmath78% over the entire period from the zero age main sequence to an age of @xmath45 billion years and only by @xmath79% from @xmath80 billion years to @xmath45 billion years .
the relation between @xmath48 and @xmath39 discussed above can be restated using the calculated time - dependence of the solar effective temperature , which is summarized in table [ tab : teff ] . in the period between @xmath81 gyr and @xmath45 gyr ,
the effective temperature is essentially constant ( to an accuracy of @xmath82% ) .
this constancy of the effective temperature results in the scaling relation @xmath83 that is valid in this period .
the effective temperature changes by an order of magnitude larger fraction during the evolution up to the present - age sun , which results in a dependence closer to @xmath84 ( cf . also figure 2 of demarque & guenther 1991 ) .
figure [ fig : efractions ] shows , for the standard model , the energy generated by different nuclear fusion reactions as a function of solar age .
the present - day total solar luminosity , @xmath85 , is the unit of luminosity in figure [ fig : efractions ] .
the branch of the @xmath31 chain that is denoted in figure [ fig : efractions ] by @xmath31 + @xmath11he-@xmath11he ( the dashed curve ) proceeds primarily through the reactions @xmath86 . for simplicity
, we include all @xmath31 reactions in this sum but do not show explicitly the @xmath87 reactions in the above scheme .
the small energy contribution due to @xmath87 reactions is included in the calculations that led to figure [ fig : efractions ] .
the @xmath11he-@xmath14he branch ( the dot - dashed curve ) includes the nuclear reactions that produce both the @xmath32be and the @xmath28b neutrinos : @xmath88 and @xmath89 .
the cno reactions are denoted by cno in figure [ fig : efractions ] .
the branch that terminates via the @xmath11he-@xmath11he reaction dominates the solar energy generation throughout the main sequence lifetime shown in figure [ fig : efractions ] , although the cno reactions overtake the @xmath11he-@xmath14he branches at an age of about eight billion years . at an age of one billion years
, @xmath90% of the solar luminosity is generated through @xmath31 reactions plus the @xmath11he-@xmath11he termination , @xmath91% through the @xmath11he-@xmath14he termination , and only @xmath54% through the cno cycle .
the situation gradually changes as the sun heats up . at the present epoch
, the @xmath11he-@xmath11he termination produces @xmath92% of the solar luminosity and the branches terminating through the @xmath11he-@xmath14he reaction generate @xmath93% of the solar luminosity .
the cno cycle produces @xmath94% of the present - epoch luminosity . by the time the standard solar model reaches an age of eight billion years , the percentages are @xmath95% , @xmath96% , and @xmath97% , respectively , for @xmath11he-@xmath11he , @xmath11he-@xmath14he , and cno reactions .
the energy loss due to gravitational energy expansion ranges from @xmath98% to @xmath99% to @xmath100% as the sun s age increases from one billion years to the present epoch to eight billion years .
figure [ fig : czvsageplus ] shows the depth of the convective zone,@xmath101 , as a function of age for the standard solar model .
correspondingly , figure [ fig : czmass ] shows the mass of the convective zone , @xmath102 , as a function of age . in both cases ,
the temporal dependence from the zero age main sequence to the current epoch is describable by a simple function of @xmath103 .
we limit the time period covered in figure [ fig : czvsageplus ] and figure [ fig : czmass ] to be less than @xmath104 billion years , since between @xmath104 and @xmath105 billion years of age semi - convection begins to influence the computed @xmath101 and @xmath102 .
the evolution of the depth of the solar convective zone was previously studied by demarque & guenther ( 1991 ) in an investigation that did not include element diffusion .
the occurrence of semi - convection in evolved solar models is discussed here for the first time , to the best of our knowledge .
the onset of semi - convection is triggered by the effects of element diffusion , which was first included in detailed solar models in the early 1990 s ( see , e.g. , bahcall & loeb 1990 ; proffitt & michaud 1991 ; bahcall & pinsonneault 1992 ) .
the process works essentially as follows .
the time scale for diffusion decreases as the surface convective zone becomes shallower ; the metal abundance increases steadily below the surface convective zone . at some point in the evolutionary history , near an age of @xmath104
billion years , the opacity from the enhanced metal abundance below the surface convective zone becomes large enough to make the metal - rich radiative layers below the surface convective zone convectively unstable .
however , the mixing of the metals into the envelope causes a local drop in the metal abundance @xmath106 and and the opacity @xmath107 at the base of the convective zone .
this result in turn causes the convective zone to recede until the metal abundance builds up again .
richer , michaud , & turcotte ( 2000 ) discuss a related phenomenon in a and f stars .
we noticed the existence of semi - convection in our standard solar model only because we made precise plots of the depth and mass of the convective zone as a function of time ( see figure [ fig : czvsageplus ] and figure [ fig : czmass ] ) .
the effects of semi - convection were not noticeable in plots of external quantities such as the solar luminosity or effective temperature ( see [ subsec : radiusetal ] ) . in our current code
, if a region is convectively unstable according to the schwarzschild criterion , then we instantly mix this material with the other material in the convective zone . a more accurate treatment , which allows for the possibility of a composition gradient being established at the base of the convective zone , is required to calculate reliably the influence of semi - convection once it begins to be perceptible in the numerical solutions [ see , for example , merryfield ( 1995 ) , canuto ( 2000 ) and references quoted therein ] .
this is why we have terminated the plots of@xmath101 and @xmath102 at a solar age of @xmath108 yr .
from the zero age main sequence to the present solar age , the depth of the convective zone is approximately proportional to the contemporary solar radius , @xmath109 , i.e. , @xmath110 the zero - age main sequence value is @xmath111 and the present - day value is @xmath112 , which corresponds to an approximately @xmath113% decrease in the depth of the convective zone over the age of the sun .
however , if we normalize the depth of the convective zone to the contemporary solar radius , the change is very small , @xmath114 and @xmath115 .
after @xmath104 billion years , @xmath116 and , in terms of the contemporary solar radius , @xmath117 .
the relative evolution between the convective zone depth and the total solar radius , @xmath118/d t = -0.002 $ ] per @xmath40 yr , is very slow . over the period from the initial age main sequence to the present epoch ,
the mass of the solar convective zone decreases with time as @xmath119 , @xmath120 to an accuracy of about @xmath121% .
the zero - age main sequence value of the mass included within the convective zone is @xmath122 , which decreases to @xmath123 at the present epoch .
however , this proportionality is no longer valid for larger ages , as can be seen in figure [ fig : czmass ] .
the mass of the convective zone is @xmath124 at @xmath104 billion years . at about @xmath125 billion years
, the previous monotonic decrease of the mass of the convective zone is reversed and @xmath102 begins to increase with time .
this behavior is not shown in figure [ fig : czmass ] since the calculated increase in the mass of the convective zone occurs in the same time frame as the onset of semi - convection .
figure [ fig : cznormalizeprop ] shows some properties at the base of the convective zone as a function of solar age .
the figure displays the time dependence of the temperature , t(cz , t ) ( solid curve ) , the density , @xmath126 ( dotted ) , the pressure , @xmath127 ( short dashes ) , and the radiative opacity , @xmath128 ( long dashes ) .
for convenience in plotting , each of the physical variables has been divided by its value at the current epoch . in cgs units , the standard model parameters have the following values at an age of @xmath129
yr : @xmath130 k , @xmath131 , @xmath132 , and @xmath133 . the temperature at the base of the convective zone decreases by @xmath113% from the zero - age main - sequence to the current solar age , i. e. , to a good approximation @xmath134 the opacity at the base of the convective zone decreases by @xmath135% over the same period .
the density and the pressure at the base of the convective zone decrease by much larger quantities , by factors of @xmath136 and @xmath137 , respectively .
equation [ eqn : tproptorminus1 ] is valid to an accuracy of about @xmath138 throughout the @xmath139 yr shown in figure [ fig : cznormalizeprop ] .
the base of the convective zone is defined by the schwarzschild criterion . because the adiabatic gradient is nearly constant in time ,
the equations of stellar structure imply that the quantity @xmath140 will be approximately constant at the base of the convective zone . from figure
[ fig : cznormalizeprop ] , we see that both the opacity and the temperature decrease slowly at the base of the convective zone .
solar models therefore compensate for the increase of the luminosity by the decrease of the pressure at the boundary between radiative and convective equilibrium .
figure [ fig : trhop ] shows the time dependence of the central values for the temperature , density , and pressure of the standard solar model .
the results are normalized to the computed values for the present epoch . over the @xmath45 billion years shown in figure [ fig : trhop ] , the central temperature increases by about @xmath141% . at the zero - age main sequence ,
the central temperature is @xmath142k and is @xmath143k at the current epoch ; the central temperature of the model reaches @xmath144k at a solar age of @xmath45 billion years .
this increase is very similar to that of the solar radius .
in fact , the central value of the temperature is , to a reasonable approximation , proportional to the solar radius , @xmath109 .
the ratio @xmath145 to an accuracy of @xmath78% ( @xmath146% ) from the zero ago - main sequence to the current epoch ( to a solar age of @xmath104 billion years ) .
the sun derives its luminosity by burning hydrogen into helium .
the hydrogen mass fraction at the center of the sun , @xmath147 , decreases from @xmath148 at the zero - age main sequence to @xmath149 at the current epoch , a decrease by more than a factor of two . at an of @xmath104 billion years
, the mass fraction has decreased to @xmath150 , a decrease by a factor of almost five from the zero - age value .
the remaining hydrogen is burned very quickly , with @xmath147 reaching @xmath151 at a solar age of @xmath45 billion years .
the @xmath0-mode oscillation frequencies change as the structure of the sun evolves with age .
since only low degree modes ( small spherical harmonic , @xmath152 ) are observable for stars other than the sun , we shall discuss the changes in just these modes .
for the high - order ( large radial node , @xmath153 ) , low - degree modes that are expected to be observed in stars , the frequencies satisfy the following relation to a good approximation ( e.g. , tassoul 1980 ) : @xmath154 here @xmath155 is the cyclic frequency of a mode of order @xmath153 and degree @xmath152 , and @xmath156 is in general a slowly varying function of frequency .
the departures from the simple relation in eq .
( [ eq : largesep ] ) and eq .
( [ eq : lplustwo ] ) provide diagnostic information about the stellar interior . in the simplest approximation ,
the modes are uniformly spaced in radial order . the so - called `` large separation , '' @xmath157 , is approximately constant , @xmath158 the large separation @xmath157 can be related to the sound speed by the formula @xmath159 in eq .
( [ eq : sepeq ] ) , the integration variable is the distance @xmath160 from the center of the star ; the range of integration is from the solar center to the surface radius @xmath161 .
the sound speed is denoted by @xmath162 . according to this equation
, @xmath157 is the inverse of the sound travel time across a stellar diameter .
thus as the the sound speed decreases with increasing solar age , the sound travel time increases and the large separations decrease .
it also follows from eq .
( [ eq : largesep ] ) that the frequencies of modes whose degrees have the same parity approximately coincide , @xmath163 the departure from eq .
( [ eq : lplustwo ] ) , the `` small frequency separation , '' is defined by the relation @xmath164 and is sensitive to conditions in the stellar core .
the small separation @xmath165 depends on the gradient of the solar sound speed , particularly near the stellar core .
the average over @xmath153 of @xmath165 is proportional to a quantity @xmath166 given by @xmath167 where @xmath168 is a suitable reference value of @xmath169 ( scherrer et al .
1983 ; christensen - dalsgaard 1988 ) .
the average small frequency separations provide a measure of the age of the star ( see , e.g. , ulrich 1986 ; christensen - dalsgaard 1988 ; guenther 1991 ) .
the solar sound speed decreases with age .
the main decrease occurs near the core , in large part due to an increase with age of the mean molecular weight as a result of hydrogen being converted to helium .
the sound - speed develops a local minimum at the center .
thus @xmath170 is positive near the center of the star ( in contrast to the negative gradient elsewhere ) , which reduces the value of the integral in eq .
( [ eq : smallcont ] ) .
the small frequency separations are more sensitive to age than the large frequency separations , although the large separations also decrease with increasing age .
table [ tab : separations ] lists for our standard solar model the calculated unweighted averages ( denoted by angular brackets ) of the large and small frequencies separations as a function of solar age .
the large separations were averaged over modes with @xmath171 , @xmath172 , @xmath80 and @xmath27 and @xmath153 between @xmath30 and @xmath173 .
this range of @xmath153 was chosen because it is the observed range for most data sets .
the set of small separations given in the third column of table [ tab : separations ] ( @xmath174 ) was calculated for @xmath171 and @xmath175 modes ; the small separations in the fourth column of table [ tab : separations ] ( @xmath176 ) were calculated for @xmath177 and @xmath178 modes .
the small separations were averaged over the same range of @xmath153 as the large separations .
lccc + age ( gyr ) & @xmath179 & @xmath180 & @xmath181 + + 0.000 & 170.57 & 19.65 & 31.34 + 0.547 & 164.03 & 17.84 & 28.76 + 1.083 & 160.22 & 16.47 & 27.55 + 1.888 & 154.87 & 15.25 & 25.08 + 2.246 & 152.62 & 14.70 & 24.29 + 2.871 & 147.98 & 13.42 & 22.29 + 3.408 & 144.52 & 12.60 & 21.10 + 4.034 & 140.16 & 11.67 & 19.68 + 4.570 & 136.10 & 10.57 & 17.97 + 5.017 & 132.87 & 9.92 & 16.95 + 5.464 & 129.57 & 8.98 & 15.53 + 5.911 & 126.19 & 8.37 & 14.50 + 6.358 & 122.82 & 7.45 & 13.09 + 6.805 & 119.27 & 6.90 & 12.07 + 7.252 & 115.75 & 6.02 & 10.65 + + 4observations + bison+lowl & 135.33 & 10.51 & 17.81 + golf & 135.12 & 10.46 & 17.75 + golf2 & 134.84 & 10.22 & 17.18 + gong & 134.86 & 10.04 & 18.24 + mdi & 134.95 & 10.14 & 17.64 + + as expected , both large and small separations decrease with time , with small separations decreasing much more rapidly than large separations .
the decrease is slightly larger for the @xmath182 separations than for the @xmath183 separations because the @xmath171 modes are more sensitive to core conditions than the @xmath177 modes .
table [ tab : separations ] also shows the results obtained from different helioseismology observations .
references for the observations are given in
[ sec : soundspeeds ] , which is where we discuss sound speeds .
we do not include in table [ tab : separations ] results for the large and small separations for lowl1 , because for this data set there are not enough low degree modes to form a robust average over the range of selected @xmath153 . the calculated values for a solar age of @xmath184 gyr should be compared with the observed values listed in table [ tab : separations ] . the calculated large separation is slightly different from the observed values , which may reflect an uncertainty in the detailed physics of the surface layers .
the uncertainty in the physics of the surface layers introduces errors in the calculated model frequencies .
these frequency - dependent errors are not completely removed in the process of calculating the large separations from the standard solar model .
the surface uncertainties are removed more successfully in the calculation of the small separations .
in this section , we describe eleven solar models , seven of which are slight variants on the theme of the standard solar model ( see [ subsec : variants ] ) and four of which are deficient in one or more significant aspects of the physics used in their construction ( see [ subsec : deviants ] ) .
nine of these solar models have been described in detail in bpb2000 , where their helioseismological properties were investigated extensively . hence the descriptions here will be brief . since the first report ( davis , harmer , & hoffman 1968 ) that the solar neutrino event rate in the chlorine experiment was less than the predicted rate ( bahcall , bahcall , & shaviv 1968 ) , obtained by using the then standard solar model and the then standard electroweak theory ,
there have been many studies of deviant solar models that were designed to ` explain ' the solar neutrino problem .
the first non - standard solar model was proposed by ezer & cameron ( 1968 ) , who suggested that the flux of @xmath28b neutrinos could be reduced if the central hydrogen abundance could be maintained near its initial value by continuous mixing in the solar core .
( for a discussion of , and references to , the extensive early work on this problem , see chapter 5 of bahcall 1989 . ) with the advent of precise measurements of solar @xmath0-modes that extend deep into the solar interior , many non - standard models have been explicitly shown to be inconsistent with the inferred solar sound speeds ( bahcall , pinsonneault , basu , & christensen - dalsgaard 1997 ) or the @xmath0-mode frequencies ( guenther & demarque 1997 ) . we explore here the range of solar parameters predicted by various non - standard models , even those that are strongly inconsistent with helioseismological data .
our purpose is to set extreme limits on predicted solar parameters , such as the luminosity evolution or neutrino emission , rather than the traditional goal of avoiding new neutrino physics .
other authors have used non - standard solar models in connection with helioseismology for a variety of different applications , including , for example , constraining the cross section for the @xmath31 interaction ( antia & chitre 1999 ) , limiting the amount of mass loss from the sun ( guzik & cox 1995 ) , and constraining the amount of anomalous energy transport by wimps ( christensen - dalsgaard 1992 ) .
turcotte & christensen - dalsgaard ( 1998a)@xcite have considered the impact of changes in heavy element abundances on the properties of some models ; the pattern of effects they found are consistent with those obtained here .
table [ tab : interior ] and table [ tab : cz ] summarize some of the important physical characteristics of the complete set of the nine solar models whose helioseismological properties we studied in bpb2000 plus two models with slightly lower values of @xmath3 that are studied here for the first time .
we present in table [ tab : interior ] for each of the models the central temperature , density , and pressure ; the helium and heavy element mass fractions in the initial model ; the helium and heavy element mass fractions in the solar center .
we give in table [ tab : cz ] the helium and heavy element abundances at the solar surface ; the mixing length parameter , @xmath156 ; and the radius and temperature at the base of the convective zone , as well as the mass included in the convective zone .
all quantities are shown for the model at the present epoch .
we will now define the eleven variant and deviant solar models .
lccccccc model&@xmath185&@xmath186&@xmath187 & @xmath188&@xmath189&@xmath190&@xmath191 + standard&15.696&152.7&2.342&0.2735&0.0188&0.6405&0.0198 + nacre&15.665&151.9&2.325&0.2739&0.0188&0.6341&0.0197 + as00&15.619&152.2&2.340&0.2679&0.0187&0.6341&0.0197 + gn93 & 15.729 & 152.9 & 2.342 & 0.2748 & 0.02004 & 0.6425 & 0.02110 + pre - m.s .
& 15.725 & 152.7 & 2.339 & 0.2752 & 0.02003 & 0.6420 & 0.02109 + rotation & 15.652 & 148.1 & 2.313 & 0.2723 & 0.01934 & 0.6199 & 0.02032 + @xmath192 & 15.729 & 152.9 & 2.342 & 0.2748 & 0.02004 & 0.6425 & 0.02110 + @xmath193 & 15.728 & 152.9 & 2.341 & 0.2748 & 0.02004 & 0.6425 & 0.02110 + no diffusion & 15.448 & 148.6 & 2.304 & 0.2656 & 0.01757 & 0.6172 & 0.01757 + old physics & 15.787 & 154.8 & 2.378 & 0.2779 & 0.01996 & 0.6439 & 0.02102 + @xmath56 & 15.621 & 153.5 & 2.417 & 0.2722 & 0.02012 & 0.6097 & 0.02116 + mixed & 15.189 & 90.68 & 1.728 & 0.2898 & 0.02012 & 0.3687 & 0.02047 + lcccccc model&@xmath194&@xmath195&@xmath156&@xmath196&@xmath197 & @xmath198 + & & & & @xmath199&@xmath200&@xmath201 + standard&0.2437&0.01694&2.04&0.7140&0.02415&2.18 + nacre&0.2443&0.01696&2.04&0.7133&0.02451&2.19 + as00&0.2386&0.01684&2.05&0.7141&0.02394&2.18 + gn93&0.2450 & 0.01805 & 2.06 & 0.7124 & 0.02457 & 2.20 + pre - m.s.&0.2455 & 0.01805 & 2.05 & 0.7127 & 0.02443 & 2.20 + rotation&0.2483 & 0.01797 & 2.03 & 0.7144 & 0.02388 & 2.15 + @xmath192 & 0.2450 & 0.01806 & 2.06 & 0.7123 & 0.02461 & 2.20 + @xmath193 & 0.2450 & 0.01806 & 2.06 & 0.7122 & 0.02467 & 2.20 + no diffusion&0.2655 & 0.01757 & 1.90 & 0.7261 & 0.02037 & 2.09 + old physics&0.2476 & 0.01796 & 2.04 & 0.7115 & 0.02455 & 2.21 + @xmath56&0.2422 & 0.01811 & 2.03 & 0.7151 & 0.02309 & 2.17 + mixed & 0.2535 & 0.01782 & 1.85 & 0.7315 & 0.01757 & 2.02 + the nacre model was constructed using the same input physics as our standard model except that we use for the nacre model the charged particle fusion cross sections recommended in the nacre compilation ( angulo et al .
we specify the fusion cross sections used for this model more fully in
[ subsubsec : nacre ] .
the model gn93 was considered our standard model in bpb2000 and differs only in the adopted value of @xmath202 ( grevesse & noels 1993 ) from the current standard model ( see [ sec : standard ] ) which has @xmath203 ( grevesse & sauval 1998 ) .
the model as00 is the same as the two models described above except that it has lower heavy element abundance @xmath204 ( asplund 2000 ) .
as a consequence of a more detailed calculation of the solar atmosphere , asplund ( 2000 ) suggests that all meteoritic abundances should be adjusted downward by 0.04 dex .
all of the models described below , in [ subsec : deviants ] as well as in this subsection , use the grevesse and noels ( 1993 ) composition mix with @xmath205 .
model pre - m.s .
is evolved from the pre - main sequence stage , but otherwise is the same as our standard model .
the model rotation incorporates mixing induced by rotation and is a reasonable upper bound to the degree of rotational mixing which is consistent with the observed depletion of lithium in the sun(pinsonneault et al .
the prescriptions for calculating this model are described in 5 of pinsonneault ( 1997 ) and in bpb2000 .
there has been considerable discussion recently regarding the precise value of the solar radius ( cf .
antia 1998 ; schou et al .
1997 ; brown & christensen - dalsgaard 1998 ) and some discussion of the effects of the uncertainty in radius on the quantities inferred from the helioseismological inversions ( cf .
basu 1998 ) .
we have therefore considered two models which were constructed with the same input physics as bp2000 , but which have model radii which differ from the radius assumed in constructing the standard model . @xmath206
has a radius of 695.78 mm , which is the radius that has been determined from the frequencies of f - modes ( cf .
antia 1998 ) and @xmath193 has a radius of 695.508 mm , which is the solar radius as determined by brown & christensen - dalsgaard ( 1998 ) , who used the measured duration of solar meridian transits during the 6 years 19811987 and combined these measurements with models of the solar limb - darkening function to estimate the value of the solar radius .
all of these variant models are approximately as consistent with the helioseismological evidence as the standard model ( see bpb2000 ) .
for example , the rms sound speed differences between the variant models and bp2000 are : @xmath207% ( pre - m.s . ) , @xmath208% ( rotation ) , @xmath209% ( radius@xmath210 ) , and @xmath207% ( radius@xmath211 ) .
the average difference ( rms ) between the four variant models and the standard model is @xmath212% .
we shall see in
[ sec : neutrinos ] that the differences predicted by these models for the important neutrino fluxes are all less than @xmath213% .
the model ` old physics ' was constructed using the old yale equation of state ( cf .
guenther et al .
1992 ) , supplemented with the debye - hckel correction ( cf .
bahcall , bahcall , & shaviv 1968 ) and older opal radiative opacities ( iglesias , rogers , & wilson 1992 ; kurucz 1991 ) .
the model includes helium and heavy element diffusion and nuclear reaction cross section data in the same way as our standard model .
the s@xmath214 model was calculated assuming that the rate of the @xmath11he@xmath215be reaction is zero , which implies that no @xmath32be or @xmath216b neutrinos are produced . in the standard solar model , about @xmath217% of the terminations of the @xmath31 chain
involve the @xmath11he@xmath215be reaction , whose rate is proportional to s@xmath218 .
the no diffusion model does not include helium or heavy - element diffusion and therefore represents the state - of - the art in solar modeling prior to 1992 ( cf .
bahcall & ulrich 1988 ; bahcall & pinsonneault 1992 ; proffitt 1994 ) .
the model mixed has an artificially mixed core , with the inner @xmath219% by mass ( @xmath220% by radius ) required to be chemically homogeneous at all times .
this model was constructed to be similar to the prescription of cumming & haxton ( 1996 ) , who changed by hand the @xmath11he abundance as a function of radius in the final bp95 ( bahcall & pinsonneault 1995 ) solar model in order to minimize the discrepancy between measurements of the total event rates in neutrino experiments and the calculated event rates .
cumming and haxton did not calculate the time evolution of their model .
we showed in bpb2000 that the mixed , no diffusion , and @xmath221 models are strongly disfavored by helioseismological data .
we use these deviant ( or deficient ) models here to test the robustness of the discrepancies between solar model predictions and solar neutrino measurements .
( for references and discussions to earlier work on these deviant models , see bahcall 1989 , christensen - dalsgaard 1995 , bahcall , pinsonneault , basu , & christensen - dalsgaard 1997 , , guenther & demarque 1997 , and cox , kidman , and newman 1985 ) .
we have already seen in figure [ fig : lumnormalized ] that the luminosity evolution predicted by solar models is essentially the same for the deviant models and for the standard solar model .
this section presents neutrino fluxes and predicted event rates in different solar neutrino experiments ; the fluxes and event rates are calculated using the eleven solar models described in [ sec : variantdeviant ] .
we also give other aspects of the solar model that influence the interpretation of solar neutrino experiments .
as described in [ subsec : standardcharacteristics ] , we present at http://www.sns.ias.edu/@xmath37jnb a detailed numerical table which gives the fraction of each of the eight important neutrino fluxes that is produced in each spherical shell .
these neutrino production fractions are important for calculating the effect of msw ( matter ) oscillations in the sun , but will not be discussed further here .
section [ subsec : fluxestoday ] presents the fluxes of electron type neutrinos that are produced in the sun according to the standard solar model and to the eight variant solar models considered in this paper .
we compare the results of these predictions with measurements of the total rates in the chlorine solar neutrino experiment , the sage and gallex + gno gallium experiments , and the kamiokande and super - kamiokande water cherenkov detectors .
we also describe in this section how we estimate the errors ( asymmetric in some cases ) on the predicted fluxes and event rates .
in [ subsec : electrondensity ] , we present detailed results for the electron number density versus solar radius .
the msw conversion of electron type neutrinos to other neutrino types depends upon the radial distribution of the electron number density . in previous presentations of the standard model
, we have not given the electron number density in the outer parts of the sun with sufficient precision to permit calculations of a subset of the currently allowed matter transitions .
section [ subsec : fluxesvsage ] presents the calculated time - dependences of the most important solar neutrino fluxes , the @xmath222 , @xmath32be , @xmath28b , and @xmath223n neutrino fluxes .
we present in [ subsubsec : standardneutrino ] the neutrino fluxes and experimental event rates predicted by the standard model and contrast these results with the observed rates . in
[ subsubsec : uncertainties ] , we describe the procedures and the ingredients used to calculate the uncertainties in the neutrino fluxes and the event rates .
we compare in
[ subsubsec : variantdeviantneutrino ] the calculated neutrino fluxes and experimental event rates for the eight variant and deviant solar models with the results for the standard model and with the measured solar neutrino event rates .
llccc source&&cl&ga&li + & & ( snu)&(snu)&(snu ) + pp&@xmath224&0.0&69.7&0.0 + pep&@xmath225&0.22&2.8&9.2 + hep&@xmath226&0.04&0.1&0.1 + @xmath227&@xmath228&1.15&34.2&9.1 + @xmath229&@xmath230&5.76&12.1&19.7 + @xmath231&@xmath232&0.09&3.4&2.3 + @xmath233&@xmath234&0.33&5.5&11.8 + @xmath235&@xmath236&0.0&0.1&0.1 + total&&@xmath237&@xmath238&@xmath239 + table [ tab : bestestimate ] gives the neutrino fluxes and their uncertainties for our standard solar model . in order to obtain the most precise values that we can for the predicted fluxes
, we have recomputed the standard model discussed elsewhere in this paper .
we use in this subsection the most recently determined absolute value for the solar luminosity , @xmath240 ( frhlich & lean 1998 , crommelynck , fichot , domingo , & lee 1996 ) , which is @xmath6% smaller than the value used in the model calculations discussed elsewhere in this paper .
the largest changes due to the @xmath6% decrease in the solar luminosity are a @xmath80% decrease in the @xmath28b neutrino flux and a @xmath172% decrease in the @xmath32be neutrino flux .
all other quantities calculated in this paper are changed by negligible amounts .
the adopted uncertainties in different input parameters are given in table 2 of bahcall , basu , & pinsonneault ( 1998 ) , which we refer to hereafter as bp98 .
we also present in table [ tab : bestestimate ] the calculated event rates in the chlorine , gallium , and lithium experiments .
the rate in the super - kamiokande experiment is usually quoted as a fraction of the best - estimate theoretical flux of @xmath28b neutrinos , assuming an undistorted ( standard ) energy spectrum .
lccc & bp2000&measured&measured / bp2000 + chlorine&@xmath237&@xmath241&@xmath242 + gallex + gno&@xmath238&@xmath243&@xmath244 + sage&@xmath238&@xmath245&@xmath246 + @xmath229-kamiokande&@xmath247$]&@xmath248$]&@xmath249 + @xmath229-super - kamiokande&@xmath250$]&@xmath251$]&@xmath252 + hep - super - kamiokande&9.3&@xmath253&@xmath254 + table [ tab : numeasurements ] compares the predictions of the combined standard model , i. e. , the standard solar model ( bp2000 ) and the standard electroweak theory ( no neutrino oscillations ) , with the results of the chlorine , gallex + gno , sage , kamiokande , and super - kamiokande solar neutrino experiments .
the observed rate in the chlorine experiment is @xmath241 snu ( lande 2000 ; davis 1994 ; cleveland et al .
1998 ) , which is to be compared to the calculated value of @xmath237 snu .
this discrepancy between calculated and observed neutrino capture rates has been approximately the same for more than three decades ( cf .
bahcall , bahcall , & shaviv 1968 ; davis , harmer , & hoffman 1968 ; bahcall 1989 ) .
the average of the sage ( abdurashitov et al .
1999 ; gavrin 2000 ) and the gallex ( hampel et al . 1999 ) plus gno ( bellotti 2000 ) results is @xmath255 snu , which is more than @xmath256 away from the calculated standard rate of @xmath238 snu .
after @xmath257 days of data acquisition , the flux of @xmath28b neutrinos measured by super - kamiokande is @xmath258 10^{-6}\,\,{\rm cm^{-2}s^{-1}}$ ] ( suzuki 2000 ; fukuda et al . 1998 ) , which corresponds to @xmath259 of the bp2000 predicted flux .
comparing the second and third columns of table [ tab : numeasurements ] , we see that the predictions of the combined standard model differ by many standard deviations from the results of the solar neutrino experiments . the flux of @xmath19 neutrinos was calculated for bp2000 using the most recent theoretical evaluation by marcucci et al .
( 2000a , b ) of the cross section factor @xmath18 , which is @xmath16 times larger than the previous best estimate .
the most recent preliminary report ( after @xmath257 days of data ) of the super - kamiokande collaboration ( suzuki 2000 ) is that the @xmath19 flux observed in their @xmath260 scattering experiment is @xmath261 times the best - estimate from bp98 .
since the bp2000 estimate of the @xmath19 flux is a factor of @xmath16 times larger than the flux quoted in bp98 , the best - estimate theoretical @xmath19 flux now agrees with the best - estimate experimental @xmath19 flux measurement , although we do not attach much significance to this agreement since we can not quote an uncertainty on the theoretical estimate ( see discussion of the @xmath19 reaction in [ subsubsec : uncertainties ] ) .
the event rates predicted by bp2000 for the chlorine , gallium , and super - kamiokande solar neutrino experiments are within two percent of the rates predicted for the bp98 standard solar models .
as far as these experiments are concerned , the effects of using the improved heavy element composition essentially cancels the effect of correcting the error in the opacity interpolation ( see [ subsec : definitionstandard ] ) .
the difference in the @xmath32be flux predicted by bp98 and bp2000 is only @xmath262% .
the @xmath32be flux will be measured by the borexino solar neutrino experiment .
we have calculated the uncertainties in the neutrino fluxes and in the experimental event rates by including the published errors in all experimental quantities and by taking account of the correlated influence of different input parameters using the results of detailed solar model calculations . the procedure for calculating
the uncertainties has been developed over the past three decades and is described in detail in chapter 7 of bahcall ( 1989 ) ( see also bahcall & pinsonneault 1992 , 1995 , and bahcall , basu , & pinsonneault 1998 ) . in order that the reader can see the specific implementation of the uncertainty calculations
, we are making available the exportable fortran code that evaluates the rates in different neutrino experiments and also calculates the uncertainties in the individual neutrino fluxes and experimental rates .
the code , exportrates.f , is available at http://www.sns.ias.edu/@xmath37jnb/sndata/sndata.html .
the uncertainties in the nuclear fusion cross sections ( except for @xmath19 , see below ) were taken from adelberger et al .
( 1998 ) , the neutrino cross sections and their uncertainties are from bahcall ( 1994 , 1997 ) and bahcall et al .
( 1996 ) , the luminosity and age uncertainties were adopted from bahcall & pinsonneault ( 1995 ) , the @xmath7 fractional uncertainty in the diffusion rate was taken to be @xmath44% ( thoul , bahcall , & loeb 1994 ) and the opacity uncertainty was determined by comparing the results of fluxes computed using the older los alamos opacities with fluxes computed using the modern livermore opacities ( bahcall & pinsonneault 1992 ) .
we follow the discussion in bahcall & pinsonneault ( 1995 ) and adopt a @xmath263 uncertainty in the heavy element abundance of @xmath264 this uncertainty spans the range of values recommended by grevesse ( 1984 ) , grevesse & noels ( 1993 ) , and grevesse & sauval ( 1998 ) over the fourteen year period covered by the cited grevesse et al . review articles .
the uncertainty adopted here is about twice as large as the uncertainty recommended by basu & antia ( 1997 ) based on their helioseismological analysis . in support of the larger uncertainty used in this paper
, we note that the difference between the grevesse & noels ( 1993 ) values of @xmath203 is @xmath263 according to eq .
( [ eqn : zoverx ] ) .
we include for the first time in this series of papers the uncertainty in the small @xmath235 neutrino flux due to the uncertainty in the measured @xmath10-factor for the reaction @xmath265 .
we use the @xmath7 uncertainty 18.1% estimated by adelberger et al .
it was an oversight not to include this uncertainty in our previous calculations . the only flux for which we do not quote an estimated uncertainty
is the @xmath19 flux ( see table [ tab : bestestimate ] ) .
the difficulty of calculating from first principles the nuclear cross section factor @xmath18 is what has caused us not to quote in this series of papers an uncertainty in the @xmath19 flux ( see discussion in bahcall 1989 and in bahcall & krastev 1998 ) .
the @xmath19 reaction is uniquely difficult to calculate among the light element fusion reactions since the one - body and two - body contributions to the reaction rate are comparable in magnitude but opposite in sign , so that the net result is sensitive to a delicate cancellation
. also , two - body axial currents from excitations of @xmath157 isobars are model dependent .
in addition , the calculated rate is sensitive to small components in the wave function , particularly @xmath166-state admixtures generated by tensor interactions .
these complications have been discussed most recently and most thoroughly by marcucci et al .
( 2000b ) .
the calculated errors are asymmetric in many cases .
these asymmetries in the uncertainties in the neutrino fluxes and experimental event rates result from asymmetries in the uncertainties of some of the input parameters , for example , the important @xmath222 , , and @xmath35n + p fusion reactions and the effect of excited states on neutrino absorption cross sections . to include the effects of asymmetric errors
, the code exportrates.f was run with different input representative uncertainties and the different higher ( lower ) rates were averaged to obtain the quoted upper ( lower ) limit uncertainties . in order to estimate the systematic uncertainties associated with different treatments of the nuclear fusion reactions ,
we have constructed a solar model that is the same as the standard model discussed in [ subsubsec : standardneutrino ] , except that we have used the charged particle fusion cross sections recommended in the nacre compilation ( angulo et al .
1999 ) rather than the fusion cross sections determined by adelberger et al .
we will refer to this solar model as the nacre model .
the low energy cross section factors , @xmath10 , that are recommended by the nacre collaboration and by adelberger et al . agree within their stated @xmath7 uncertainties for all of the fusion reactions that are important for constructing a solar model .
the only important solar nuclear reactions for which the nacre collaboration did not recommend interaction rates are the electron capture reactions that produce the @xmath32be and the @xmath87 neutrinos ; the nacre collaboration also did not provide energy derivatives for the cross section factors of the cno reactions . wherever the data necessary for computing solar fusion rates was not available in the nacre compilation , we continued to use the adelberger et al .
( 1998 ) recommended values in computing the nacre model .
llccc source&&cl&ga&li + & & ( snu)&(snu)&(snu ) + pp&@xmath266&0.0&69.8&0.0 + pep&@xmath267&0.22&2.8&9.1 + hep&@xmath268&0.04&0.1&0.1 + @xmath227&@xmath269&1.15&34.5&9.1 + @xmath229&@xmath270&6.20&13.1&21.2 + @xmath231&@xmath271&0.08&2.9&2.1 + @xmath233&@xmath272&0.28&4.7&10.3 + @xmath235&@xmath273&0.0&0.1&0.1 + total&&@xmath274&@xmath238&@xmath275 + table [ tab : nacre ] gives the calculated neutrino fluxes and capture rates predicted by the nacre solar model . in all cases ,
the fluxes for the nacre solar model agree with the fluxes calculated with the standard solar model to well within the @xmath7 uncertainties in the standard model fluxes .
the @xmath32be flux from the nacre model is @xmath172% larger than for the standard model and the @xmath28b flux is @xmath45% higher .
the chlorine capture rate predicted by the nacre model is @xmath213% higher than for the standard model ; the predicted rates for the nacre model and the standard model differ by less than @xmath172% for the gallium and lithium experiments .
we conclude that this estimate of the systematic uncertainties due to the relative weights given to different determinations of nuclear fusion cross sections suggests likely errors from this source that are significantly smaller than our quoted @xmath7 errors for the standard model neutrino fluxes and event rates ( cf .
table [ tab : bestestimate ] ) .
similar conclusions have been reached by morel , pichon , provost , & berthomieu ( 1999 ) in an independent investigation ( see also castellani et al .
1997 ) .
lcccccccccc & pp&pep&hep&@xmath227&@xmath229&@xmath231 & @xmath233&@xmath235&cl&ga + & ( e10)&(e8)&(e3)&(e9)&(e6)&(e8)&(e8)&(e6)&(snu)&(snu ) + standard&5.96&1.40&9.3&4.82&5.15&5.56&4.88&5.73&7.7&129 + nacre&5.97&1.39&9.4&4.85&5.54&4.93&4.24&5.39&8.1&129 + as00&5.99&1.41&9.4&4.62&4.70&5.25&4.56&5.33&7.1&126 + gn 93 & 5.94 & 1.39 & 9.2 & 4.88 & 5.31 & 6.18 & 5.45 & 6.50 & 8.0 & 130 + pre - m.s . &
5.95 & 1.39 & 9.2 & 4.87 & 5.29 & 6.16 & 5.43 & 6.47 & 7.9 & 130 + rotation & 5.98 & 1.40 & 9.2 & 4.68 & 4.91 & 5.57 & 4.87 & 5.79 & 7.4 & 127 + @xmath192 & 5.94 & 1.39 & 9.2 & 4.88 & 5.31 & 6.18 & 5.45 & 6.50 & 8.0 & 130 + @xmath193 & 5.94 & 1.39 & 9.2 & 4.88 & 5.31 & 6.18 & 5.45 & 6.50 & 8.0 & 130 + no diffusion & 6.05 & 1.43 & 9.6 & 4.21 & 3.87 & 4.09 & 3.46 & 4.05 & 6.0 & 120 + old physics & 5.95 & 1.41 & 9.2 & 4.91 & 5.15 & 5.77 & 5.03 & 5.92 & 7.8 & 130 + @xmath56 & 6.40 & 1.55 & 10.1 & 0.00 & 0.00 & 6.47 & 5.64 & 6.70 & 0.8 & 89 + mixed & 6.13 & 1.27 & 6.2 & 3.57 & 4.13 & 3.04 & 3.05 & 3.61 & 6.1 & 115 + table [ tab : fluxesnine ] compares the calculated neutrino fluxes for the seven variant solar models described in [ subsec : variants ] and for the four deficient solar models described in [ subsec : deviants ] with the fluxes obtained for the standard solar model .
the range of the @xmath28b neutrino fluxes among the seven standard - like models ( rows 17 of table [ tab : fluxesnine ] ) is only @xmath276% , a factor of two or three smaller than the uncertainty ( due to other sources ) in the calculated @xmath229 flux of the standard model ( see table [ tab : bestestimate ] ) .
the spread in @xmath227 flux is @xmath277 ; the range in @xmath278 and @xmath279 rates is @xmath280 snu and @xmath281 snu , respectively .
the deviant models listed in table [ tab : fluxesnine ] are all deficient in some important aspect of the physics used in their calculation .
the no diffusion , old physics , @xmath56 , and mixed models all give such bad agreement with helioseismology ( compared to the first seven models ) that the comparison with the data can not be made on the same scale as for the standard - like models ( see bpb2000 ) .
the old physics model gives a rms difference between the calculated and measured sound speeds that is more than twice as large as when the standard model is used .
for the @xmath56 model , the rms discrepancy is seven times worse in the solar interior than for the standard model ; for the no diffusion model , the disagreement is about seven times worse averaged over the sun than for the standard model .
the mixed model is the worst of all ; the rms disagreement in the solar core is about @xmath220 times larger than for the standard solar model .
even if one is willing to consider solar models that predict sound speeds which deviate so drastically from the measured helioseismological values , the four deviant solar models do not describe well the solar neutrino data . for the old physics , no diffusion , and mixed solar models , the predicted rates for the gallium experiments lie between @xmath282 snu and @xmath283 snu , which is to be contrasted with the observed rate of @xmath284 snu , which is at least an @xmath285 discrepancy . for the @xmath286 model ,
the predicted rate in the chlorine experiment is @xmath287 snu , which is also about eight standard deviations less than the observed value , @xmath288 snu .
the probability of converting an electron type neutrino to a muon or tau neutrino in the sun depends upon the profile of the electron number density as a function of solar radius . for particular values of the electron density , neutrino energy , and neutrino mass ,
neutrinos can be resonantly converted from one type of neutrino to another .
the mikheyev - smirnov resonance occurs if the electron density at a radius @xmath160 satisfies @xmath289 where @xmath290 is the electron number density measured in @xmath291 , @xmath292 is avogadro s number , @xmath293 is the neutrino mixing angle in vacuum , @xmath294 is the absolute value of the difference in neutrino masses between two species that are mixing by neutrino oscillations , and @xmath295 is the neutrino energy .
ccccc @xmath296&@xmath297&&@xmath296&@xmath297 + 0.01 & 2.008e+00 & & 0.55 & @xmath2981.527e@xmath29801 + 0.05 & 1.956e+00 & & 0.60 & @xmath2983.605e@xmath29801 + 0.10 & 1.827e+00 & & 0.65 & @xmath2985.585e@xmath29801 + 0.15 & 1.662e+00 & & 0.70 & @xmath2987.428e@xmath29801 + 0.20 & 1.468e+00 & & 0.75 & @xmath2989.098e@xmath29801 + 0.25 & 1.249e+00 & & 0.80 & @xmath2981.099e+00 + 0.30 & 1.012e+00 & & 0.85 & @xmath2981.330e+00 + 0.35 & 7.687e@xmath29801 & & 0.90 & @xmath2981.642e+00 + 0.40 & 5.269e@xmath29801 & & 0.95 & @xmath2982.164e+00 + 0.45 & 2.914e@xmath29801 & & 1.00 & @xmath2986.806e+00 + 0.50 & 6.466e@xmath29802 + figure [ fig : ne ] and table [ tab : ne ] give the electron number density as a function of solar radius for the standard solar model ( bp2000 ) .
a much more extensive numerical file of the electron number density versus radius is available at http://www.sns.ias.edu/@xmath37jnb ; this file contains the computed values of the electron number density at @xmath299 radial shells .
we see from figure [ fig : ne ] that for typical values of the neutrino parameters that allow the so - called lma and sma msw solutions which fit all of the currently available solar neutrino data ( e.g. , bahcall , krastev , & smirnov 1998 ) , the experimentally most important @xmath28b neutrinos ( @xmath300 ) satisfy the resonance condition , eq .
( [ eq : mswresonance ] ) , at radii that are smaller than the radius of the convective zone . for the so - called low msw solutions and for all msw solutions with @xmath301 , the resonance radius falls in the outer part of the sun .
we have not previously published accurate values for the electron density in the outer parts of the sun , @xmath302 .
the straight line in figure [ fig : ne ] is an approximation to the electron number density in the standard solar model of bahcall & ulrich ( 1988 ) ( see bahcall 1989 ) .
this approximation , which has been used for over a decade by different groups analyzing solar neutrino data , is @xmath303 figure [ fig : ne ] shows that the approximation given in eq .
( [ eq : nestraight ] ) fails badly in the outer regions of the sun .
recently , several different analyses have been published using the electron number density shown in figure [ fig : ne ] ( or , more precisely , the computer file for @xmath304 which is available at http://www.sns.ias.edu/@xmath37jnb ) .
the effective density of particles for interacting with sterile ( right handed ) neutrinos is not the electron number density discussed in the previous subsection , but rather @xmath305 ( see mikheyev & smirnov 1986 ; lim & marciano 1988 ; barger et al . 1991 ) , where @xmath306 and @xmath307 is the number density of neutrons . since nearly all of the neutrons in the sun are either in @xmath14he or in heavier elements with @xmath308 , it is easy to derive an analytic expression that relates @xmath305 and @xmath290 . one obtains @xmath309 figure [ fig : nsterile ] and table [ tab : nsterile ] give the radial distribution of @xmath305 in the standard solar model .
the functional form of @xmath310 is similar to the function form of @xmath304 .
the straight line in figure [ fig : nsterile ] is given by an equation of the same form as eq .
( [ eq : nestraight ] ) that describes @xmath304 except that the coefficient for @xmath311 is @xmath312 ( instead of @xmath313 for @xmath314 ) .
ccccc @xmath296&@xmath297&&@xmath296&@xmath297 + 0.01 & 1.885e+00 & & 0.55 & -1.901e-01 + 0.05 & 1.853e+00 & & 0.60 & -3.978e-01 + 0.10 & 1.757e+00 & & 0.65 & -5.956e-01 + 0.15 & 1.611e+00 & & 0.70 & -7.777e-01 + 0.20 & 1.425e+00 & & 0.75 & -9.436e-01 + 0.25 & 1.209e+00 & & 0.80 & -1.133e+00 + 0.30 & 9.731e-01 & & 0.85 & -1.364e+00 + 0.35 & 7.303e-01 & & 0.90 & -1.676e+00 + 0.40 & 4.887e-01 & & 0.95 & -2.198e+00 + 0.45 & 2.536e-01 & & 1.00 & -6.839e+00 + 0.50 & 2.701e-02 + the number density @xmath305 is about @xmath220% smaller than the electron number density , @xmath290 , in the center of the sun , where helium is most abundant . in the central and outer regions of the sun , @xmath315 is about @xmath316% less than @xmath290 .
since the slopes of the straight - line fitting functions are the same in figure [ fig : ne ] and figure [ fig : nsterile ] , the neutrino survival probabilities are the same for sterile and for active neutrinos as long as the adiabatic approximation is valid ( see , e.g. , 9.2 of bahcall 1989 ) .
figure [ fig : fluxesvsage ] shows the most important solar neutrino fluxes , the @xmath222 , @xmath32be , @xmath28b , and @xmath223n fluxes , as a function of solar age .
the fluxes displayed in the figure were computed using the standard model and are normalized by dividing each flux by its value at the present epoch , labeled by the word ` today ' in the figure ( cf .
10 of guenther and demarque 1991 for a similar figure plotted on a logarithmic scale ) .
the @xmath222 flux is relatively constant over the entire @xmath45 billion years show in figure [ fig : fluxesvsage ] . at the beginning of its lifetime ,
the @xmath222 flux is about @xmath317% of its current value and reaches @xmath318% of its present value after @xmath319 billion years . at the current epoch ,
the flux is changing very slowly , about @xmath81% per billion years .
the @xmath222 flux reaches a maximum , @xmath81% larger than its current value , at a solar age of @xmath320 billion years and then declines slowly and steadily to @xmath321% of its present value at an age of @xmath45 billion years .
the @xmath32be and @xmath28b neutrino fluxes increase monotonically and by larger amounts than the @xmath222 flux .
both the @xmath32be and the @xmath28b fluxes begin with very low fluxes relative to their current values , @xmath113% and @xmath27% , respectively , of their intensities at @xmath322 yr . at a solar age of @xmath45 billion years , the @xmath32be neutrino flux is @xmath319 times larger than it is today and the @xmath28b neutrino flux is @xmath323 times larger than today . at the current epoch
, the @xmath32be flux is increasing by about @xmath324% per billion years and the @xmath28b flux is increasing faster , about 120% per billion years . the @xmath223n neutrino flux has the most interesting time dependence . in the first @xmath2 y on the main sequence ,
the @xmath223n flux is much larger than its current value because @xmath34c has not yet been burned to the equilibrium value appropriate for the cno cycle .
the reaction @xmath325n occurs relatively often in this early stage of solar evolution and the neutrino flux from @xmath223n beta - decay has a peak value of about @xmath326 times its current flux .
the minimum @xmath223n flux , @xmath327% of its present value , is attained at a solar age of @xmath328 billion years .
thereafter , the @xmath223n flux increases steadily as the central temperature of the solar model increases and reaches an intensity of @xmath329 times its current value at a solar age of @xmath45 billion years .
section [ subsec : panoramic ] presents a panoramic view of the predicted standard model sound speeds and compares the observations and the calculations on a scale that is relevant for interpreting solar neutrino experiments . in [ subsec : zoom ] , we compare on a zoomed - in scale the standard model calculations with the results of six helioseismological measurements .
the zoomed - in scale used in this subsection can reveal fractional discrepancies between calculations and observations that are smaller than @xmath24% . in
[ subsec : rotation ] , we consider a particular solar model that includes rotation in a plausible way .
this model smoothes the composition discontinuity at the base of the convective zone , which locally improves the agreement with the measured sound speeds but worsens the overall rms agreement ( within the uncertainties allowed by input data ) . for neutrino emission ,
the predictions of the rotation model are not significantly different from the standard model predictions .
some authors ( see , e.g. , guenther & demarque 1997 ) compare their solar models directly with the @xmath0-mode oscillation frequencies rather than with inverted quantities such as the sound speed . the reader is referred to the guenther and demarque paper for a discussion of the direct comparison method , its application , and additional references .
we have chosen to use the sound speed profile because the inversion process that produces the inferred sound speeds allows one to remove the uncertainties , common to all @xmath0-mode oscillation frequencies , that arise from the near - surface regions of the sun .
these common uncertainties are due to the treatment of convection , turbulence , and non - adiabatic effects .
inversion techniques are designed to minimize the effects of these outer layers ( see , e.g. , basu , christensen - dalsgaard , hernandez , and thompson 1996 ) moreover , the sound speed profile summarizes in a robust way the results obtained for many thousands of oscillation frequencies .
finally , the inversion procedure allows one to isolate different regions of the sun , which is important in the context of discussions regarding solar neutrinos .
the neutrinos are produced deep in the solar interior . in bpb2000 , we have discussed in detail the systematic uncertainties and assumptions related to the inversion for the sound speeds and for the less - accurately determined density .
figure [ fig : diffbp00best ] shows the fractional differences between the calculated sound speeds for the standard model and what may be the most accurate available sound speeds measured by helioseismology , the lowl1 + bison measurements presented in basu et al .
these sound speeds are derived from a combination of the data obtained by the birmingham solar oscillation network ( bison ; cf .
chaplin et al . 1996 ) and the low-@xmath330 instrument ( lowl ; cf .
tomczyk et al .
1995a , b ) .
the rms fractional difference between the calculated and the measured sound speeds is @xmath331 over the entire region in which the sound speeds are well measured , @xmath332 . in the solar core , @xmath333 ( in which about @xmath334% of the solar energy and neutrino flux are produced in a standard solar model )
, the rms fractional difference between measured and calculated sound speeds is @xmath335 .
the standard model sound speeds agree with the measured sound speeds to @xmath24% whether or not one limits the comparison to the solar interior or averages over the entire sun .
systematic uncertainties @xmath336 are contributed to the sound speed profile by each of three sources : the assumed reference model , the width of the inversion kernel , and the measurement errors ( see bpb2000 ) .
the vertical scale of figure [ fig : diffbp00best ] was chosen so as to include the arrow marked `` @xmath337 ) . ''
this arrow indicates the typical difference between solar model speeds and helioseismological measurements that would be expected if the discrepancy between the gallium solar neutrino measurements and the predictions in table [ tab : bestestimate ] were due to errors in the solar physics of the standard solar model ( see discussion in bp98 ) .
figure [ fig : soundspeed ] and table [ tab : soundspeeds ] give the sound speeds versus the solar radius that are calculated using the standard solar model .
the sound speed declines from about @xmath338 km / s in the solar core to about @xmath339 km / s at @xmath340 .
an extensive table of the standard model sound speeds is available at http://www.sns.ias.edu/@xmath37jnb .
ccccc @xmath161&@xmath341&&@xmath161&@xmath341 + @xmath199&(100 km / s)&&@xmath199&(100 km / s ) + 0.01 & 5.057 & & 0.50 & 2.990 + 0.05 & 5.101 & & 0.55 & 2.817 + 0.10 & 5.078 & & 0.60 & 2.651 + 0.15 & 4.870 & & 0.65 & 2.483 + 0.20 & 4.550 & & 0.70 & 2.295 + 0.25 & 4.210 & & 0.75 & 2.035 + 0.30 & 3.898 & & 0.80 & 1.761 + 0.35 & 3.627 & & 0.85 & 1.476 + 0.40 & 3.389 & & 0.90 & 1.162 + 0.45 & 3.179 & & 0.95 & 0.778 + figure [ fig : differences6 ] compares the results of six precise observational determinations of the solar sound speed with the results of our standard solar model .
the vertical scale has been expanded by a factor of @xmath342 relative to figure [ fig : diffbp00best ] in order to show the small but robust discrepancies between the calculations and the observations and to indicate the size of the differences between the various measurements . in the deep solar solar interior where neutrinos are produced , @xmath343 , the differences between the various observational determinations of the sound speed are comparable to the differences between bp2000 and any one of the measured sets of sound speeds
. the @xmath344mode frequencies used in deriving the observed sound speeds shown in figure [ fig : differences6 ] were obtained from a number of different sources .
in addition to the lowl1 + bison data described in
[ subsec : panoramic ] , we have used data from a number of other sources . ( 1 ) data from the first year of lowl observations .
the sound speed inversions are described in basu et al .
( 1997 ) and are referred to as lowl1 in this paper .
( 2 ) data from the michelson doppler imager ( mdi ) instrument on board the solar and heliospheric observatory ( soho ) during the first 144 days of its operation ( cf .
rhodes et al .
the results of the sound speed inversions using these data are from basu ( 1998 ) .
( 3 ) the frequencies obtained from the data obtained by the global oscillation network group ( gong ) between months @xmath81@xmath113 of its observations .
the solar sound - speeds are from basu ( 1998 ) .
( 4 ) initial observation taken by the global oscillations at low frequencies ( golf ) instrument on board soho , combined with intermediate data from mdi .
we have labeled the sound speeds obtained as golf1 and the solar sound speed results can be found in turck chize et al .
( 5 ) more recent data from golf ( thiery et al . 2000 ) , combined with intermediate degree data obtained from the first 360 days observations by the mdi instrument ( schou et al .
the sound - speed results are described in basu et al .
these results have been labeled as golf2 .
there are other helioseismological data sets that could have been used . of these ,
the data by toutain et al .
( 1998 ) may be the most relevant .
this data set has been discussed extensively in basu et al .
toutain et al .
( 1998 ) were the first to show the effect of line - profile asymmetry on low - degree modes .
the sound - speed inversions in their paper were , however , very different from what had been obtained previously .
basu et al .
( 2000 ) showed that this difference results primarily from the two modes , @xmath175 , @xmath345 and @xmath175 @xmath346 , whose frequencies ( and errors on the frequencies ) were suspect . in a later paper , bertello et al .
( 2000 ) confirmed most of the low frequency modes in the toutain et al .
( 1998 ) data set , but not the two questionable modes . additionally , they determined the frequencies of some very low - frequency low - degree modes . the sound - speed inversion of bertello et al .
( 2000 ) is similar to the inversion obtained in basu et al .
( 2000 ) and in the present paper ; the errors in the inversions are also similar .
the similarity in the errors may appear somewhat surprising at first glance given that the bertello et al .
frequencies have very low errors , but the similarity in the overall uncertainties is explained by the fact that the inversion errors are determined by the error in the bulk of the frequencies ( which are mdi intermediate-@xmath152 frequencies ) , and not just the errors in the low - degree modes .
the mdi and gong sets have good coverage of intermediate degree modes .
the mdi set has p - modes from @xmath347 up to a degree of @xmath348 while the gong set has modes from @xmath347 up to @xmath349 .
however , both these sets are somewhat deficient in low degree modes .
the lowl1 + bison combination , on the other hand , has a better coverage of low degree modes , but has modes from @xmath347 only up to @xmath350 .
the golf data sets only contain low - degree modes ( @xmath351 ) and hence have to be combined with other data before they can be used to determine the solar sound speed profile .
figure [ fig : differences5 ] shows the somewhat less precise agreement that was obtained between bp98 ( bahcall , basu , & pinsonneault 1998 ) and the helioseismological data . the bp98 model and
the observed sound speeds agree in the solar interior to about the same accuracy as for bp2000 and the observed speeds . however , the bp98 model sound speeds are about @xmath24% smaller than the observed speeds in the broad intermediate region between @xmath352 and @xmath353 ( cf .
figure [ fig : differences6 ] and figure [ fig : differences5 ] ) . averaged over the entire region over which good measurements are available , @xmath354 to @xmath355 , the rms fractional difference between the bp98 model and the lowl1 + bison sound speeds is @xmath356 , which sound be compared with a rms difference of @xmath331 for the bp2000 model ( see [ subsec : panoramic ] ) , that are significant for helioseismology : the correction of the opacity interpolation error and the slightly different heavy element mixture adopted for the bp2000 model .
for helioseismology , the only significant different between gn93 and bp98 is the correction of the interpolation error ; the heavy element abundances used for both models are the same ( with @xmath202 ) .
when a figure like figure [ fig : differences6 ] was constructed using gn93 instead of bp2000 , the general form of the small differences was practically the same .
in fact , the rms difference averaged over the whole sun between gn93 and lowl1 + bison is @xmath357 , which is slightly better than the value of @xmath358 for bp2000 versus lowl1 + bison .
one therefore obtains the correct impression by comparing figure [ fig : differences6 ] and figure [ fig : differences5 ] . ] . in bp98
, we speculated that the broad feature of disagreement at the @xmath24% level in figure [ fig : differences5 ] might be due to a combination of small errors in the adopted radiative opacities or in the equations of state .
we investigated the implications for solar neutrino fluxes of the possibility that the small , broad discrepancy was due to an opacity error and concluded that if this were the case then the corrected solar model would predict @xmath32be and @xmath28b neutrino fluxes that are about @xmath213% larger than the fluxes predicted by bp98 . indeed , the origin of the broad discrepancy is due to an error in interpolating the radiative opacity near the edges of the opacity tables , as explained in [ subsec : definitionstandard ] .
the effect on the neutrino fluxes is somewhat smaller than we had estimated , an increase ( relative to bp98 ) of @xmath80% for @xmath32be neutrinos and @xmath27% for @xmath28b ( cf .
the results for gn93 in table [ tab : bestestimate ] and table 1 of bp98 ) .
figure [ fig : differences6rot ] compares the sound speeds of the rotation model with the six precise observed sets of sound speeds . in the region between @xmath359 and @xmath360 ,
the agreement between the rotation model and the helioseismological data is slightly less good than with the standard solar model ( compare figure [ fig : differences6 ] and figure [ fig : differences6rot ] ) .
quantitatively , over the entire region between @xmath354 to @xmath340 , the rms fractional difference between the rotation model and the lowl1 + bison sound speeds is @xmath361 , which is to be compared with @xmath362 for bp2000 . in the solar core ( @xmath363 ) , the two models , rotation and bp2000 , have almost identical rms fractional differences with respect to the lowl1 + bison sound speeds , @xmath364% and @xmath365% , respectively .
figure [ fig : differencesbprot ] compares both the rotation and the standard model with just the lowl1 + bison sound speeds , this figure shows that the rotation model gives marginally better agreement with the measured sound speeds right at the base of the convective zone , comparable agreement in the deep interior , @xmath366 , and slightly less good agreement in the intermediate region between @xmath359 and @xmath360 .
the neutrino fluxes calculated with the rotation model lie well within the estimated errors in the standard model fluxes , as can be seen easily by comparing the fluxes and the errors given in table [ tab : bestestimate ] and table [ tab : fluxesnine ] .
the @xmath32be flux for the rotation model is @xmath27% less than the standard model @xmath32be and the @xmath28b flux is @xmath213% lower than the corresponding standard model value .
the rotation model predicts a capture rate by @xmath278 that is 0.3 snu less than the standard model rate and a @xmath279 capture rate that is 2 snu less than the standard model rate .
table [ tab : yandcz ] gives the calculated present - epoch values for the depth of the convective zone , r(cz ) , and the surface helium abundance , @xmath194 , for all @xmath326 of the solar models considered in this paper .
lcc & @xmath194&@xmath367 + standard&0.244&0.714 + nacre&0.244&0.713 + as00&0.239&0.714 + gn93&0.245&0.712 + pre - ms&0.246&0.713 + rotation&0.248&0.714 + @xmath368&0.245&0.712 + @xmath369&0.245&0.712 + no diffusion&0.266&0.726 + old physics&0.248&0.712 + @xmath56&0.242&0.715 + mixed&0.254&0.732 + measured&@xmath370&@xmath371 + the observed values determined using measurements of @xmath0-modes are @xmath372 ( basu & anita 1995 ; for earlier work see christensen - dalsgaard , gough , & thompson 1991 ) and @xmath373 ( basu & anita 1997 ; see also richard et al .
1996 ) .
the quoted errors are systematic ; the statistical uncertainties are much smaller .
for example , the cited uncertainty for @xmath194 is designed to span the two values obtained when using two different equations of state ( cf .
basu & anita 1997 ) .
there is no rigorous way of establishing a confidence level based upon agreement within , for example , one or two times the estimated systematic uncertainty .
we point out that the recent reassessment of @xmath3 by grevesse and sauval ( 1998 ) resulted in a value of @xmath203 which differs from the helioseismologically recommended value of @xmath374 by twice the quoted uncertainty of the helioseismological determination . as a rule - of - thumb
, we shall regard agreement within two times the quoted systematic uncertainty as ` satisfactory ' and agreement within one times the quoted systematic uncertainty as ` excellent ' .
nine of the eleven solar models considered in this paper give excellent or satisfactory agreement with the observed depth of the convective zone .
the only exceptions to the good agreement with the measured convective zone depth are the no diffusion and the mixed models , which are both strongly disfavored by the helioseismological measurements .
the no diffusion , @xmath56 , and as00 models are the only ones that are not within twice the quoted uncertainty in the measured surface helium abundance .
the no diffusion model yields a surface helium abundance that is more than five times the quoted uncertainty away from the helioseismological measurement , which very strongly disfavors the no diffusion model .
the surface helium abundance calculated for the as00 model is also rather far from the helioseismological value , more than three times the quoted uncertainty in the helioseismological determination .
this discrepancy should be examined more fully in future years as the abundance determinations are refined and the helioseismological determination of the helium abundance is repeated and the input data to the helioseismological analysis is varied over a wide range of allowed possibilities .
this paper provides new information about four topics : 1 ) the characteristics of the standard solar model at the current epoch ( see [ subsec : discusscurrent ] below ) ; 2 ) time dependences of important characteristics of the standard solar model ( [ subsec : discusstime ] ) ; 3 ) neutrino fluxes and related quantities for standard and variant solar models ( [ subsec : neutrinocharacteristics ] ) ; and 4 ) measured versus calculated solar sound speeds ( [ subsec : discusssoundspeeds ] ) .
extensive numerical data that are useful for applications are available at http://www.sns.ias.edu/@xmath37jnb .
we present detailed numerical tabulations of the computed characteristics of our standard solar model , which is defined and discussed in [ sec : standard ] .
these tables include , as a function of the solar radius , the enclosed mass fraction , the temperature , mass density , electron number density , the pressure , and the luminosity fraction created in a given spherical shell , as well as the mass fractions of @xmath375h , @xmath11he , @xmath14he , @xmath32be , @xmath34c , @xmath35n , and @xmath36o . over the years , previous numerical versions of our standard model have been used for a variety of purposes that range from comparisons with other stellar evolution codes , estimating the importance in the sun of newly considered physical effects , searching for possible instabilities in the sun , comparison with helioseismological measurements , and the calculation of processes ( especially the msw effect ) that influence the propagation of solar neutrinos . in the past , we have published in hard copy form increasingly more detailed and precise numerical tables of the characteristics of the solar interior .
the capabilities of current calculations and the requirements of some of the most interesting applications have made complete hard copy publication no longer appropriate .
we have therefore limited ourselves in
[ sec : standard ] to describing briefly the ingredients we use in calculating the current standard model .
we present the numerical results in exportable data files that are available at http://www.sns.ias.edu/@xmath37jnb . the total luminosity in the standard model increases by @xmath376% from the zero - age main sequence stage to the present epoch . over the same period ,
the effective temperature varies by only @xmath377% .
these predictions constitute constraints on models for the evolution of the earth .
the predicted time evolution of the solar luminosity is robust .
figure [ fig : lumnormalized ] shows that all solar models , even those models with deficient physics that are strongly disfavored by helioseismological measurements , predict essentially the same luminosity evolution .
the average rms deviation of the deviant models , the mixed , no diffusion , and @xmath56 models , from the standard solar model luminosity is only @xmath172% over the history of the sun from @xmath172 gyr to the current epoch ( see [ subsec : radiusetal ] for more details ) .
we have presented in [ sec : timedependences ] the time evolution of some of the principal physical quantities characterizing the solar core ( the central temperature , density , pressure , and hydrogen mass fraction , as well as the fractions of the solar luminosity generated by different nuclear reactions ) .
we also present the evolution of important quantities at the base of the convective zone ( radiative opacity , temperature , density , and pressure ) .
we hope that these data and the scaling relations we have inferred will be sufficient to permit a future physical understanding of the time dependences using analytic and semi - analytic arguments .
we find some simple scaling relations .
for example , the solar luminosity , @xmath63 , is approximately related to the solar radius , @xmath109 , as @xmath378 .
the depth of the convective zone , @xmath101 , scales as @xmath379 , and the central temperature , @xmath380 , shows a similar behavior , @xmath381 .
moreover , we find that the mass of the convective zone , @xmath102 satisfies @xmath382 .
the effective temperature is approximately constant , varying by only @xmath383 from a solar age of 2 billion years to 8 billion years .
these results make predictions that are potentially testable . in principle , the measurement of the luminosity ( by astrometry ) of a star with the same mass and chemical composition ( @xmath3 ) as the sun would allow the prediction of the star s effective temperature , the depth and mass of the convective zone , and the large and small separations of the @xmath0-mode frequencies ( see for example monteiro , christensen - dalsgaard , & thompson 2000 and christensen - dalsgaard 1997 ) . in practice , it is difficult to make measurements sufficiently accurately to make possible precise tests of stellar evolution theory . for an appraisal of both the potential and the difficulty of making such measurements ,
the reader is referred to the recent papers by guenther and demarque ( 2000 ) and morel , provost , lebreton , thvenin , & berthomieu ( 2000 ) on the binary pair of approximately solar mass stars , @xmath156 centauri ab ( @xmath384 and @xmath385 ) .
figure [ fig : ne ] and table [ tab : ne ] give the electron number density as a function of position in the sun for the standard solar model .
the distribution of the electron density is required to compute the probability for matter - induced oscillations between active neutrinos .
similarly , figure [ fig : nsterile ] and table [ tab : nsterile ] give the radial distribution of the number density of scatterers of sterile neutrinos , @xmath305 , in the standard solar model .
we have not previously published precise values for the electron number density , or of @xmath305 , in the outer regions of the sun .
the outer regions are relevant for large mixing angle neutrino oscillations with relatively low neutrino mass differences ( @xmath386 ) .
table [ tab : bestestimate ] presents the neutrino fluxes and the event rates in the chlorine , gallium , lithium , and electron - scattering neutrino experiments that are predicted by the standard solar model .
these predictions assume that nothing happens to solar neutrinos after they are produced .
the table also gives estimates of the uncertainties in the fluxes and the event rates ;
[ subsec : fluxestoday ] contains a discussion of the physical origin of the uncertainties , as well as the software used to calculate the asymmetric error estimates .
how do the predictions of solar neutrino event rates compare with experiment ?
table [ tab : numeasurements ] compares the predictions of bp2000 with the results of the chlorine , gallex + gno , sage , kamiokande , and super - kamiokande solar neutrino experiments .
this table assumes nothing happens to the neutrinos after they are created in the sun .
the standard predictions differ from the observed rates by many standard deviations .
because of an accidental cancellation , the predicted solar neutrino event rates for bp2000 and bp98 are almost identical ( see [ subsubsec : standardneutrino ] ) .
table [ tab : fluxesnine ] compares the neutrino fluxes and the experimental event rates for all nine of the solar models whose helioseismological properties were investigated in bpb2000 , plus two additional standard - like models considered here which have somewhat different heavy - element to hydrogen ratios .
the seven standard - like models ( the first seven models in table [ tab : fluxesnine ] ) all produce essentially the same neutrino predictions ; the spreads in the predicted @xmath222 , @xmath32be , and @xmath28b fluxes are @xmath79% .
@xmath387% . and
@xmath388% , respectively .
the calculated rates for the seven standard - like solar models have a range of @xmath280 snu for the chlorine experiment and @xmath281 snu for the gallium experiments .
the estimated total errors from external sources ( see table [ tab : bestestimate ] ) , such as nuclear cross section measurements and heavy element abundances , are about a factor of three larger than the uncertainties resulting from the solar model calculations .
we have investigated one possible source of systematic errors , the relative weights assigned to different determinations of nuclear fusion cross sections .
we calculated the neutrino fluxes and predicted event rates using the nacre ( angulo et al .
1999 ) fusion cross sections rather than the adelberger et al .
( 1998 ) cross sections .
the nacre parameters lead to slightly higher predicted event rates in solar neutrino experiments .
however , all changes in the neutrino fluxes and event rates between the nacre - based predictions and the standard predictions ( based upon adelberger et al . nuclear parameters ) are much less than the @xmath7 uncertainties quoted for the standard model ( cf . table [ tab : bestestimate ] and table [ tab : nacre ] ) .
the neutrino event rates predicted by all seven of the standard - like solar models considered here are inconsistent at the @xmath389 level ( combined theoretical and experimental errors ) with the results of the two gallium experiments , gallex and sage , assuming no new physics is occurring .
the inconsistency with the chlorine experiment is similar but more complex to specify , since the largest part of the theoretical uncertainty in the calculated standard capture rate is due to the electron - type neutrinos from @xmath28b beta - decay .
the fractional uncertainty in the @xmath28b flux depends upon the magnitude of the flux created in the solar interior . moreover , the amount by which this @xmath28b flux is reduced depends upon the adopted particle physics scenario .
a similar level of inconsistency persists even for the ad hoc deficient models , such as the @xmath56 and mixed models , that were specially concocted to minimize the discrepancy with the neutrino measurements .
for example , the calculated rates for the mixed model are @xmath390 snu for the chlorine experiment and @xmath391 snu for gallium experiments .
the mixed model is @xmath392 below the measured chlorine rate and @xmath393 below the measured gallium rate .
in addition , the deficient models are strongly disfavored by the helioseismology measurements .
figure [ fig : fluxesvsage ] shows the calculated time dependence of the @xmath222 , @xmath32be , @xmath28b , and @xmath223n solar neutrino fluxes . at the current epoch , the @xmath222 flux is increasing at a rate of @xmath81% per billion years , and the @xmath32be , @xmath28b , and @xmath223n fluxes at the rates of @xmath394% , @xmath318% , and @xmath395% per billion years , respectively .
since the age of the sun is estimated to be uncertain by only @xmath396 billion years ( see the appendix by wasserburg in bahcall & pinsonneault 1995 ) , the age of the sun does not represent a significant uncertainty for solar neutrino predictions .
figure [ fig : diffbp00best ] shows the excellent agreement between the helioseismologically determined sound speeds and the speeds that are calculated for the standard solar model .
the scale on this figure was chosen so as to highlight the contrast between the excellent agreement found with the standard model and the two orders of magnitude larger rms difference for a solar model that could reduce significantly the solar neutrino problems .
one would expect characteristically a @xmath316% rms difference between the observations and the predictions of solar models that significantly reduce the conflicts between solar model measurements and solar model predictions .
averaged over the entire sun , the rms fractional difference is only @xmath397% between the standard solar model sound speeds and the helioseismologically - determined sound speeds .
the agreement is even better , @xmath398 , in the interior region in which the luminosity and the neutrinos are produced .
table [ tab : soundspeeds ] presents numerical values for the sound speeds predicted by the standard solar model at representative radial positions in the sun .
all eight of the standard - like solar models considered in this paper give acceptable agreement with the measured depth of the convective zone and the surface helium abundance ( see table [ tab : yandcz ] ) .
of the twelve standard , standard - like , variant , and deviant solar models considered here , only the no diffusion and mixed models disagree strongly with the measured convective zone depth .
there are small but robust discrepancies between the measured and the calculated solar sound speeds .
figure [ fig : differences6 ] shows the fractional differences between the standard model sound speeds and the speeds measured in each of six different determinations ( using @xmath0-mode data from lowl1 , bison , gong , mdi , golf , and golf2 ) .
the vertical scale for figure [ fig : differences6 ] is expanded @xmath342 times compared to the vertical scale of figure [ fig : diffbp00best ] . as can be seen from figure [ fig : differences6 ] ,
the differences between observed and measured sound speeds are comparable over much of the sun to the differences between different measurements of the sound speed , but there is a clear discrepancy near the base of the convective zone that is independent of which observational data set is used .
there may also be a less prominent discrepancy near @xmath399 .
the agreement between the bp2000 sound speeds and the measured values is improved over what was found earlier with the bp98 model .
this improvement may be seen by comparing figure [ fig : differences6 ] and figure [ fig : differences5 ] .
the improvement is striking in the region between @xmath400 and @xmath401 and is due to the correction of an error in the interpolation algorithm for the radiative opacity ( cf .
discussion in
[ subsec : definitionstandard ] ) .
the rotation model includes a prescription for element mixing that was designed to explain the depletion of lithium .
the calculated differences between the rotation and the standard models represent a reasonable upper limit to the effects that rotation , sufficient to explain lithium depletion , might produce .
the overall agreement between the sound speeds of the rotation model and the helioseismologically determined sound speeds is slightly worse , @xmath402% rather than @xmath397% , than the agreement obtained with the standard solar model .
the differences between the solar neutrino fluxes predicted between the rotation model and the standard model is typically about @xmath403 , where @xmath404 represents the uncertainties given in table [ tab : bestestimate ] for the standard model neutrino fluxes .
we conclude that further improvements of the theoretical calculations motivated by refinements of @xmath344 mode oscillation measurements are unlikely to significantly affect the calculated solar neutrino fluxes . @xmath405 the simple relations as a function of time between the solar radius , the solar luminosity , the depth of the convective core , and the mass of the convective core ( see eq .
[ eqn : lrproportional]eq .
[ eqn : tcoverr ] ) .
we are grateful to many colleagues in solar physics , nuclear physics , and particle physics for valuable discussions , advice , criticism , and stimulation .
jnb is supported in part by an nsf grant # phy-0070928 .
bellotti , e. , et al .
( gno collaboration ) 2001 , in the proc . of the xix international conference on neutrino physics & astrophysics , sudbury , canada , june 1621 , 2000 ( amsterdam : elsevier science publishers ) , in press gavrin , v. ( sage collaboration ) 2001 , in the proc . of the xix international conference on neutrino physics & astrophysics , sudbury , canada , june 1621 , 2000 ( amsterdam : elsevier science publishers ) in press mikheyev , s. p. , & smirnov , a. yu .
1986 , in 86 massive neutrinos in astrophysics and in particle physics , proc . of the sixth moriond workshop , ed .
o. fackler , & trn thanh vn ( gif - sur - yvette : editions frontires ) , 355 | we calculate accurate solar models and report the detailed time dependences of important solar quantities .
we use helioseismology to constrain the luminosity evolution of the sun and report the discovery of semi - convection in evolved solar models that include diffusion .
in addition , we compare the computed sound speeds with the results of @xmath0-mode observations by bison , golf , gong , lowl , and mdi instruments .
we contrast the neutrino predictions from a set of eight standard - like solar models and four deviant ( or deficient ) solar models with the results of solar neutrino experiments . for solar neutrino and for helioseismological applications ,
we present present - epoch numerical tabulations of characteristics of the standard solar model as a function of solar radius , including the principal physical and composition variables , sound speeds , neutrino fluxes , and functions needed for calculating solar neutrino oscillations . |
one of the most promising `` smoking - gun '' signals for establishing the existence of the dark matter particle is the possible observation of a sharp cosmic @xmath0-ray line from dark matter annihilation or decay @xcite .
the forthcoming cherenkov telescopes @xcite , the current fermi large area telescope @xcite and the hess instrument @xcite will allow to probe this possibility with further sensitivity . if such a signal is observed in the near future , the question of the identification of the dm particle that could have caused it will become crucial . such a signal could be induced through annihilation , coannihilation or decay . for all these scenarios , it is generally assumed that the photon is emitted through the loop of a charged particle . beside this general class of models , there exist other ways along which dm could emit monochromatic photons .
one possibility consists in assuming that the @xmath0-ray line is due to a @xmath2 chern - simons interaction @xcite .
another possibility , much less studied , would be to consider a photon directly emitted by the dm particle .
this is a priori perfectly possible if dm is not exactly neutral , but is millicharged . for an annihilation such
a possibility is not much of an option because the associated @xmath0-ray line would be in general suppressed with respect to the total cross section , by the square of the millicharge . given the constraints there are on the total cross section ( in particular from the relic density in the thermal freezeout scenario ) , this would lead to a signal sizeably smaller than present or near future sensitivities .
instead for a decay , there is a priori more freedom because the decay lifetime is not so directly constrained by the relic density . in this work we consider such a decay possibility . in the following
we will first consider the two main frameworks that can in a simple way justify a millicharge for the dm particle , kinetic mixing and stueckelberg scenarios .
in such scenarios , in order to justify that the dm particle would have a slow decay , we assume that its stability is due to an accidental symmetry that , being accidental , would be naturally broken by any uv physics . along these lines , the decay is naturally slow because suppressed by powers of the uv scale , just as expected for the proton .
the appropriate language to consider in a model - independent way the possibility of a slow decay is therefore the one of the higher - dimensional operator effective theory .
unlike for an annihilation , the use of an effective theory for a decay is fully justified since one expects a clear scale separation .
consequently , such an effective theory allows for a systematic study of possibilities .
we will determine all dimension - five and dimension - six operators that can lead to a two - body radiative decay from a millicharged fermion , scalar or vector dm particle .
these operators come in addition to the effective operators which can lead to a @xmath0-ray line in the case where dm would be exactly neutral , given and studied in ref .
the former operators involve a covariant derivative of the millicharged field , whereas the latter ones can involve a photon only from the presence of a hypercharge or @xmath3 field strength @xmath4 in the operator . in the following
, we will perform a detailed analysis of the constraints that hold on the various `` millicharged operators '' for the fermionic dm case .
the scalar and vector cases will be discussed more briefly before concluding .
a simple constraint that turns out to be relevant in some cases is that the dm particle lifetime should be larger than the age of the universe .
another one concerns the emission of cosmic rays ( cr ) that could be associated to the one of the photon , either from the particle that accompanies the photon in the decay final state , or from other decays that the effective operator unavoidably predicts on top of the radiative one .
gauge invariance in particular predicts decays where the photon is replaced by a @xmath5 .
if the electromagnetic coupling to the @xmath5 is not millicharge suppressed , the flux of cosmic rays produced is much larger than the flux of monochromatic photons .
in particular , if the particle accompanying the photon in the final state is a neutrino , which is the only standard model ( sm ) particle possibility ( a decay of special interest being `` poly - monochromatic '' , i.e. monochromatic for both types of cosmic rays that are the less affected while propagating ) , we will see that an observable @xmath0-ray line is not an option , unless the dm mass is quite low . therefore , except for this case , the possibilities we will find point towards multi - component dm scenarios .
other constraints are related to the fact that along the stueckelberg scenario the dm particle is charged under a new @xmath6 gauge group , which may be at the origin of the unsuppressed emission of the associated @xmath7 .
a millicharge for a particle can either be postulated as just so ( from assuming an hypercharge such that @xmath8 is small ) or induced from a dynamical process , typically a small mixing parameter between the sm hypercharge gauge boson and a new @xmath6 gauge boson .
the first option requires another particle to carry just so the same millicharge in such a way that the dm particle can decay into it . for the more appealing second option
, one can point out two simple scenarios , depending on whether the @xmath6 gauge boson is massless or massive . in the massless case ,
a millicharge is induced for an originally neutral particle if the unbroken @xmath6 gauge boson kinetically mixes with the hypecharge gauge group @xcite , @xmath9 applying first a non - unitary transformation to get rid of this non - canonical kinetic term , one can always in a second step rotate both gauge boson fields with a unitary transformation because both gauge bosons are massless .
there is therefore some arbitrariness in defining both fields .
we make the convenient choice to go to the basis where the state which is essentially the hypercharge gauge boson couples to both @xmath10 and @xmath11 generators , whereas the other one , which is essentially the @xmath6 gauge boson , couples only to the @xmath11 generator .
it allows to put the kinetic mixing suppression in the production decay process rather than in the detection , see e.g. ref .
@xcite . in this basis , and after electroweak symmetry breaking , the covariant derivative @xmath12 becomes @xmath13 with @xmath14 . a field with charges @xmath15 , @xmath16 , and @xmath11 couples to the photon field @xmath17 with charge @xmath18 with @xmath19 . in particular
, a field with @xmath20 acquires a millicharge @xmath21 .
note that everywhere in the following we will make the approximation @xmath22 .
existing constraints on the parameters apply in general on the millicharge @xmath23 , rather than on @xmath24 directly , see below .
a value of @xmath25 is therefore not excluded .
however , it is generally expected smaller than one .
for instance , if we consider the minimal scenario where the only dm couplings are those of eq.([lagrangian_kin_mix ] ) , the thermal relic abundance of the dm is provided by the annihilation into dark photons . by requiring the right dark matter abundance
, we get a value for @xmath26 as a function of @xmath27 this constraint together with eq .
( [ mc_bound ] ) below gives the bound @xmath28 , justifying our approximation .
it is well known that if the @xmath6 symmetry is spontaneously broken , so that the corresponding gauge boson becomes massive , a kinetic mixing interaction does not induce any millicharge for an originally neutral field . in the massive case there
exists nevertheless the stueckelberg option .
the stueckelberg mechanism allows to have a massive gauge boson without breaking the corresponding gauge symmetry .
we will here consider an extension of the sm by a @xmath6 as in @xcite .
this model contains a scalar which has stueckelberg couplings to both @xmath29 and @xmath6 . as a consequence , the neutral gauge bosons mix , and in the mass eigenstates basis the covariant
derivative reads @xmath30 where @xmath31 and @xmath32 stand for the sine and cosine of the various angles with @xmath33 , with @xmath34 the `` bare '' mass of the @xmath35 gauge boson , respectively @xcite .
the expression of the electromagnetic charge is @xmath36 , with @xmath37 . in this way ,
an originally neutral field acquires a charge of @xmath38 .
note that the stueckelberg scenario as origin of a millicharged dm ( like the `` just - so '' scenario ) might be questioned by considerations of quantum gravity / string theory @xcite .
the list of possible radiative decays that could be generated by the millicharge of a particle is extremely reduced and in this sense points towards a rather precise kind of scenario . for the fermion dm case ,
there is only one decay possibility ,
@xmath39 with @xmath40 and @xmath41 necessarily carrying the same millicharge . in the following , when establishing the list of operators that could lead to a sizable monochromatic photon signal , we will not specify the exact nature of the fermionic partner of the dm in these operators . it could be either a dirac or a majorana fermion , and it could be either a particle beyond the sm or a neutrino .
the former option points towards a multi - component fermion dm scenario . note that the results obtained below , in particular those of fig .
[ fig : kin_mix ] , do not depend on how the abundancies of these components pile up to saturate the observed value of @xmath42 , except those depending on the direct detection constraints on a millicharge .
up to dimension six , there is only a very limited number of operators that can induce a @xmath39 decay from the millicharge of @xmath40 and @xmath41 .
first of all , we only find a single dimension - five operator @xmath43 for this operator and all operators below , the addition of its hermitian conjugate is implicit
. do not give any radiative two - body decays because this would require that the scalar field has both a vev and a millicharge , which would give a mass to the photon .
similarly , operators with a @xmath44 or @xmath45 do not give any radiative decays as can be seen from the use of the equations of motion .
note also that operators with an additional @xmath46 are redundant since both fermions in the operator are different fields ( i.e. it can always be reabsorbed in the definition of one of the fermion field ) . ]
the presence of the @xmath47 implies that this operator can be rewritten as a sum of operators where both covariant derivatives have been replaced by a sum of field strengths of the gauge boson to which the particle couples ( each one multiplied by the corresponding gauge coupling ) .
dipole operators ( as relevant for example for the @xmath48 decay ) , with the difference that the photon would here be radiated by a millicharged particle instead of a charged lepton or charged gauge boson . ] as for the dimension - six operators we find only three possibilities , @xmath49 their structure are equivalent up to one operator that does not produce monochromatic photons but can give other decays ( including two - body decays ) , hence different amounts of cosmic rays .
as for the operator of eq .
( [ ofdpsisig ] ) , the covariant derivative of the operators of eqs .
( [ o6fdpsisig ] , [ o6fddmsig ] ) can be traded for a sum of field strengths . in summary ,
up to dimension six , we are left with four operator structures only , as given in eq .
( [ ofdpsisig ] ) and eqs .
( [ o6fdpsisig])-([o6fd2sig ] ) . at the two - body decay level ,
the scalar field in the last three operators can intervene only through its vev . for the quantum numbers of these fields
there is in principle an infinity of possibilities and we will see how , when considering the constraints on the various operators , a simple global picture can emerge despite of this fact .
as for the non - millicharged operators in ref .
@xcite , there are a priori essentially two main ways to constraint the operators and thus to possibly discriminate between them , from @xmath0-ray line spectral features and from the associated continuum of cosmic rays produced .
the fact that the lifetime of the dm particle must be longer than the age of the universe provides an additional constraint which is relevant in special cases . by spectral features
we mean the number of @xmath0-ray lines produced , their relative energies and relative intensities .
however , for the millicharged operators there is no way to get more than one @xmath0-ray line from a unique given operator since in the final state , on top of the photon , one can only find the @xmath41 particle . to determine what are the possibilities to distinguish among the various operators from a cosmic ray multi - channel analysis , we will proceed as for the case of non - millicharged operators in ref .
the whole issue is that for a given operator , due to gauge invariance , there always is a continuum flux of cosmic rays associated to the production of a @xmath0-ray line , especially if its energy is larger than the @xmath5 boson mass .
the decay rate of the dm particle into photons is proportional to its millicharge squared .
therefore , to determine the upper bounds existing on the photon over cosmic rays ratios , we will need to know ( in some cases ) what are the bounds which hold on the millicharge of a metastable particle within the mass range we consider , @xmath50 gev . in this mass range , there are no relevant accelerator constraints @xcite .
there are nevertheless stringent constraints from cosmology as well as from direct detection data . as well - known , in the usual @xmath51cdm model , which fits well both the cmb anisotropy and large scale structure data , it is assumed that there is no dm - baryon interaction other than gravitational .
an additional dm - baryon interaction such as the one provided by a millicharge modifies this picture by rendering dm effectively `` baryonic '' .
this affects the cmb power spectrum as well as the baryon acoustic oscillations , leading to the upper bound @xcite @xmath52 where @xmath53 is the dm - baryon momentum - transfer cross - section and @xmath54 is the dm - baryon relative velocity . in our case
, the relevant cross - section is the rutherford one and @xmath55 eq .
( [ cmb_bound ] ) translates into the following bound on the dm millicharge @xmath56 where @xmath57 is the electromagnetic fine structure constant .
the direct detection bounds are much more model - dependent than the cmb ones .
they crucially depend on the mass of the particle exchanged between the nucleon and the dm particle . in the massless @xmath6 gauge boson case , the elastic nucleon - dm scattering is proportional to the inverse of the recoil energy squared , @xmath58 .
this results in a huge enhancement of the scattering cross section with respect to the usual wimp case where the cross section is typically suppressed by the inverse of the square of the gev - tev mass of the particle exchanged . for @xmath59 few gev this results in upper bounds on the millicharge of order of @xmath60-@xmath61 , see fig .
9 of ref .
@xcite ( where the @xmath62 parameter is nothing but @xmath63 in our notation ) . for the stueckelberg case ,
these constraints remain also valid as long as the @xmath7 is lighter than @xmath64 mev , see for example fig.1 of ref .
@xcite . beyond these values ,
the upper bound on the millicharge scales as @xmath65 , so that it quickly becomes weaker than the cmb bound ( this occurs for @xmath66 gev ) . in the following ,
when the bound on the millicharge is relevant to determine the upper bound on the @xmath0-ray line , we use the cmb bound .
we will see that this bound already excludes an observable @xmath0-ray line for these cases ( hence a fortiori when the direct detection bounds are stronger , this is even more excluded ) .
note also that all these direct detection constraints assume a standard local dm density , which might not apply depending on how large the dm particle millicharge is , because it can be shielded by galactic magnetic fields @xcite . for instance , when @xmath67 , the depletion of the local dm relic density from magnetic shielding begins to be sizeable , therefore weakening the direct detection bounds .
the photon over cosmic rays ratio in both kinetic mixing and stueckelberg scenarios is given by @xmath68 @xmath69 here , the @xmath70 ratios hold for the number of cosmic rays ( of a given type and of a given energy ) produced per particle p , and @xmath71 are functions of the @xmath72 and @xmath73 masses . for @xmath74 , they always are equal to unity except for the operator of eq .
( [ o6fd2sig ] ) , see below . in these equations @xmath75 is the millicharge which , as said above , is equal to @xmath76 and @xmath77 in the kinetic mixing and stueckelberg cases , respectively . as for the coupling to the @xmath5 , @xmath78 , as eqs .
( [ lagrangian_kin_mix])-([lagrangian_stuck ] ) show , it takes the simple form @xmath79 respectively ( where @xmath80 indicates the tangent of @xmath81 ) .
the coupling to the @xmath7 , which applies only in the stueckelberg case , is @xmath82 finally , the coupling to the @xmath83 , @xmath84 , can take very different values as a function of the multiplets considered in the various operators ( and the associated clebsch - gordan coefficients ) . in practice , we will consider the cases which , among all possible multiplet configurations up to @xmath3 quintuplets , minimize @xmath84 , hence maximize the @xmath85 ratio . before coming to the constraints which hold for the various operators , note that in the following we will not consider in many details the amount of cosmic rays the @xmath7 could give . the limits on the intensity of the @xmath0-ray line we will give below hold for the case where the @xmath7 does not give any cosmic rays .
this can be the case for example if the @xmath7 decays essentially to @xmath86 ( if @xmath87 ) .
this situation gives the maximum upper bound that could be reached .
when neglecting the contribution of the @xmath7 to the cosmic rays , both stueckelberg and kinetic mixing scenarios give rise to the same fig .
[ fig : kin_mix ] bounds at the lowest order in @xmath88 and @xmath24 respectively . at the end of sec .
[ constraints ] , we will discuss how our results might be affected if , instead , the @xmath7 mainly decays into sm particles .
for the unique dimension - five operator , the quantum numbers of @xmath40 and @xmath41 are necessarily the same , and in particular @xmath90 . the simple crucial remark to be done at this stage , not only valid for this operator but for all operators , is that unless the fields to which the covariant derivative applies is a singlet of both @xmath3 and @xmath29 , there will always be a two - body decay production of a @xmath5 and/or a @xmath83 in a way which is not suppressed by the value of the millicharge . as a result in this case
, the production of cosmic rays is boosted , by the inverse of the millicharge squared , with respect to the production of the @xmath0-ray .
z bosons are produced in this way as soon as the @xmath15 of the particle to which the covariant derivative applies is different from zero , see eqs .
( [ czkm ] ) and ( [ czstu ] ) . as for the @xmath84 coupling , for the dimension - five operator one can write it as @xmath91 . for an @xmath3 multiplet of dimension @xmath92 ,
@xmath93 if @xmath94 and @xmath95 in all other cases .
similarly , @xmath96 if @xmath97 and @xmath98 in all other cases . as a result , both @xmath99 and @xmath100 , hence @xmath84 , can be zero only if @xmath101 , that is to say if one considers a sm singlet . from this discussion
there are three general cases * . in this case
there is no @xmath83 production and the @xmath5 production involves two powers of the millicharge , as for the @xmath0 production . as a result
we get an unsuppressed ratio @xmath102 at the lowest order in @xmath24 for the kinetic mixing case and in @xmath88 for the stueckelberg case . in the latter case , this is a good approximation if @xmath103 , which is what is expected if there is a big splitting between the sm gauge bosons and the @xmath7 masses . * . in this case
, there is production of @xmath83 bosons and @xmath104 in eq .
( [ ratio_ddpsidpsi ] ) , meaning that @xmath105 at the lowest order in @xmath106 or @xmath88 . here
, in order to obtain a conservative model - independent upper bound , we made the hypothesis that the charged @xmath41 components produced in two - body decays together with @xmath83 bosons do not yield an important contribution to cosmic rays production . * . in this case
we have both unsuppressed production of @xmath5 and @xmath83 . here
we will consider only the case with @xmath107 and @xmath108 , as it is the one which maximizes the @xmath109 ratio of eq .
( [ ratio_ddpsidpsi ] ) .
it gives @xmath110 with @xmath111 at the lowest order in @xmath106 or @xmath88 . in this case , the relevant quantum numbers are those of @xmath41 .
the maximal ratios as a function of the value of @xmath113 follow the same pattern as for the previous operator . in the same way ,
one has three cases * : prediction ( [ ratio_a ] ) . * : prediction ( [ ratio_b ] ) . * : prediction ( [ ratio_c ] ) .
now , the relevant quantum numbers are those of the dm particle . here , the three cases are : * : prediction of eq .
( [ ratio_a ] ) . * : the ratio is maximized with @xmath115 , which gives is obtained for a situation with @xmath40 a triplet , and @xmath41 and @xmath88 quintuplets .
any combination of smaller multiplets gives a bigger value of @xmath84 .
for example , taking @xmath40 as a hyperchargeless triplet , and both @xmath41 and @xmath88 as doublets of hypercharge 1 , gives @xmath116 .
] @xmath117 * : the ratio is maximized with @xmath118 and @xmath119 , which gives @xmath120 the phenomenology of this operator is more involved than the one of the operators above because it depends in a complicated way on the couplings of both @xmath41 and @xmath40 to the various bosons .
nonetheless , maximizing the @xmath109 ratio requires @xmath41 and @xmath40 to have the same quantum numbers , which greatly reduces the complexity of the @xmath109 ratio . in the case where both @xmath40 and @xmath41 are sm singlets ,
this ratio is given by @xmath122 at the lowest order in @xmath24 or @xmath123 in the case where @xmath40 and @xmath41 are not sm singlets , the predictions are quite lengthy and we only give them for the cases where @xmath124 and @xmath125 in the appendix . unlike all previous cases where the dependence on @xmath126 is negligible for @xmath127 , here , from the longitudinal @xmath83 contribution ,
there are terms in @xmath128 which imply a power law dependence on @xmath129 , see appendix .
this is associated to the fact that , unlike the other operators , this one is not equivalent to a single operator involving a field strength .
this will give rise to a different behaviour of the constraints below , as the bounds obtained for the @xmath121 operator get stronger when @xmath129 increases .
[ fig : kin_mix ] shows the constraints obtained on the photon over cosmic rays ratios for the different operators involving a millicharged dm . for low masses , up to @xmath130 tev ,
the most stringent bounds are provided by pamela measurements of cosmic antiprotons @xcite , whereas for higher masses , the relevant constraints come from measurements of the diffuse gamma background from fermi - lat @xcite and hess @xcite .
the methodology used to obtain these constraints is the same as in ref .
@xcite . by comparing the constraints from cosmic rays on @xmath131 and the limits from direct searches for photon line spectral features ,
this plot shows which operator are compatible with a possible near future observation of a @xmath0-ray line .
a clear pattern emerges from these results . except for the case
in which the dm is a singlet of the sm gauge groups , none of the effective operators associated to a millicharged dm taken individually would be able to produce a @xmath0-ray line strong enough to meet the current experimental sensitivity without overproducing antiprotons and diffuse photons . actually , in all the cases in which the dm is not a sm singlet , the observation of a spectral photon line would be associated to an excess of cosmic rays from five to ten orders of magnitude higher than the present experimental bounds from pamela or fermi - lat and hess .
this result is interesting because it singles out a unique possibility for the sm quantum numbers of the dm : a sm singlet .
unfortunately , it does not allow to discriminate , neither among the mechanisms responsible for the millicharge of the dm , nor among the various operators . when the dm is a singlet , all operators lead to the prediction of eq .
( [ ratio_a ] ) , @xmath132 , except op.([ratio_f ] ) which differs at dm masses close to @xmath133 , but only very slightly . as fig .
1 shows , this prediction gives the maximum @xmath0-ray line intensity allowed by cosmic rays constraints .
this absolute bound turns out to be the same as in ref .
@xcite for the neutral dm scenarios .
as for a non - sm - singlet dm , in fig . 1 we have only considered the quantum numbers that maximize @xmath131 .
the maximum ratios turn out to be identical for operators @xmath89 and @xmath134 .
they differ by less than one order of magnitude for @xmath114 , which is within the uncertainty coming from the propagation models .
the only operator providing very different bounds for non - singlet dm is the operator @xmath121 , due to the dependence of the @xmath131 ratios on the dm mass .
not only are these bounds exhibiting a different behaviour for increasing @xmath129 , but they also differ by more than two orders of magnitude from all the previous bounds for @xmath135 tev . therefore , if a line were to be detected with a sensitivity of direct searches for monochromatic photons improved by several orders of magnitude , it would in principle be possible to discriminate this particular operator from the other ones where the dm is not a sm singlet .
but , in practice , this does not appear at all to be a realistic option because this basically means a @xmath0-ray line with intensity smaller than the intensity of the photon continuum observed .
putting together the results obtained for a millicharged dm and those derived for a neutral dm @xcite , we find that if a line were detected at the current experimental sensitivity without any excess of cosmic rays , it would not be possible to discriminate the millicharged sm - singlet scenario from the neutral dm case giving the same eq .
[ ratio_a ] ratio ( prediction `` a '' in @xcite ) .
however , if on the contrary a strong line were to be detected with a sizeable associated cosmic rays excess , only the more suppressed @xmath0-ray predictions `` b''-``e '' in ref .
@xcite for neutral dm could explain it .
as said above , all the bounds obtained have been computed under the hypothesis that the stueckelberg @xmath7 that could be present in two - body decay final states ( if kinematically allowed ) does not produce any cosmic rays by decaying subsequently .
if this does not hold , the operators will give rise to more suppressed bounds . to estimate how important this contribution could be ,
we consider as an example , with @xmath136 , a @xmath7 boson which decays mainly to @xmath137 , a channel which is known to produce many cosmic rays .
when the dm is not a singlet , we find bounds that are stronger by approximately two orders of magnitude .
this stems from the fact that , in this case , a single @xmath7 produces a comparable amount of cosmic rays with respect to the sm gauge bosons , but , with @xmath138 , its coupling to the dm particle is stronger than those of the @xmath5 or @xmath139 bosons .
if , instead , the dm particle is a sm singlet , the @xmath140 ratio does not depend anymore on the value of @xmath141 , but the coupling to the photon is suppressed by @xmath142 . instead of having a fixed bound on the ratio as given by eq .
( [ ratio_a ] ) , the following limit is obtained : @xmath143 the @xmath88 angle is constrained by the measurement of the @xmath5 width from lep , @xmath144 @xcite .
this decreases the bound of eq .
( [ ratio_a ] ) by three to four orders of magnitude .
interestingly , the observation of a @xmath0-ray line with intensity of order the present experimental sensitivity would therefore probe this possibility , see fig . 1 . as mentioned above for the massless kinetic mixing case , the direct detection constraints on the millicharge are stronger than the `` universal '' cmb constraints we have used for fig . [
fig : kin_mix ] . as a result , in this case , an observable @xmath0-ray line for the non - sm singlet scenario is even less of an option , not only because this would give even more cosmic rays , but also because it would give a dm lifetime smaller than the age of the universe . concerning possible rapid decays , note also that since the radiative decays are suppressed in all cases ( singlet case included ) by a factor of the millicharge squared , if there exist other ( non - radiative ) operators induced at the scale @xmath51 which destabilize the dm particle in a way which is not suppressed by this factor , they could easily induce decays much faster than the radiative ones .
these decays could induce cosmic ray fluxes above the ones observed or even give a dm lifetime shorter than the age of the universe .
the scenario is therefore viable if there is no such operator or if the associated decays do exist , but with a lifetime larger than the age of the universe and without an excessive associated production of cosmic rays .
so far , we have discussed the constraints on the intensity of a @xmath0-ray line which hold from the bounds on the millicharge itself , as this is the parameter entering in the various rates . however ,
if one has some prejudice about the values of the parameters responsible for the millicharge , there are cases where stronger bounds on the @xmath0-ray line intensity apply .
this is relevant in particular for the kinetic mixing case if , as for any gauge charge in the sm , one considers a value of the `` dark charge '' @xmath141 not far from unity .
the issue here is that in the massless hidden gauge boson case , the dm particle does not only decay to @xmath145 but it also does to @xmath146 . as the @xmath147 does not produce any cosmic rays , the latter decay is not relevant for fig . 1 .
however , it is naturally faster than the former decay by a factor of @xmath148 .
one has therefore to make sure that the resulting lifetime is not shorter than the age of the universe . imposing that @xmath149 , and imposing that the millicharge satisfies the cmb bound of eq .
( [ mc_bound ] ) , we show in fig .
2 the upper limits which hold on the intensity of the @xmath0-ray line for various values of the dark charge @xmath141 . clearly ,
for large values of @xmath141 one gets competitive bounds with respect to those derived from @xmath0-ray line direct searches , whereas smaller values give irrelevant bounds .
2 also shows the constraints we get assuming the direct detection bounds mentioned above ( i.e. fig .
9 of ref .
@xcite , disregarding possible weakening of these bounds from magnetic shielding ) .
these constraints are quite stringent , leading to unobservable @xmath0-ray lines , unless the dark charge is small enough . in fact
, @xmath150 is necessary in order to get @xmath151 .
these constraints are obtained assuming that the relic density of @xmath40 is the observed one . for smaller relic density values
one gets weaker constraints . in the stueckelberg case ,
all these considerations become obviously irrelevant as soon as the @xmath7 gauge boson mass is above @xmath129 ( or above @xmath152 gev for what concerns the direct detection bounds ) .
from the above results it is clear that , since the neutrino is not a sm singlet , it is not an option within the @xmath154 gev range considered above . moreover in this case
, since the millicharge of both fermions in the various operators must be equal , and since the millicharge of the neutrinos is extremely well bounded , the millicharge of the dm particle is also extremely well constrained .
the most stringent constraint applies on the electronic neutrino .
assuming charge conservation in @xmath155 decay , and using the experimental results from @xcite and @xcite : @xmath156 and @xmath157 , the constraint @xmath158 is obtained .
independent , less stringent upper bounds also hold from neutrino magnetic dipole moment searches , see _
e.g. _ @xcite . in the case of @xmath159 and @xmath160 ,
the most stringent constraints come from stellar evolution @xcite .
if neutrinos acquire a millicharge , their electromagnetic interactions would provoke extra energy losses in the core of red giants .
this would delay the time of helium ignition , and as a consequence , the core of red giants would be heavier than in the standard case when helium lights up .
but the mass of the red giant core at helium ignition is constrained by measurements from globular clusters .
these constraints turn into the following bound on the charge of neutrinos : @xmath161 .
this bound holds as long as @xmath162 kev , implying that it applies to all flavours . to get a lifetime allowing for an observable @xmath0-ray line , the huge suppression due to this millicharge could in principle
be compensated by considering smaller @xmath51 scales , taking typically @xmath163 gev for dimension - five operators , and @xmath164 gev for dimension - six operators . however , since the @xmath5 emission rate is enhanced with respect to the @xmath0 emission by a factor of @xmath165 , this would clearly imply a decay rate to @xmath166 leading to a lifetime much shorter than the lifetime of the universe .
similarly this would have given a huge amount of cosmic rays . in fig .
[ fig : kin_mix ] and rescale them by a factor of @xmath167 where @xmath168 refers to the bound of eq .
( [ mc_bound ] ) .
] therefore , due to several reasons , a line observed at the present sensitivities with energies above the @xmath5 mass could not be attributed at all to a millicharged dm decaying in neutrino and photon through one of the operators under study .
more generally , one could ask whether this possibility is also excluded for lower dm masses . here , there are a priori two directly connected constraints which must be fulfilled , giving an upper and a lower bound on @xmath129 .
the upper bound comes from the fact that , even if for @xmath169 the @xmath5 can not be produced on its mass shell , it can be produced off - shell and subsequently decay to a pair of fermions .
this could result in a lifetime shorter than the age of the universe and/or to too many cosmic rays .
for instance , supposing that a @xmath0 line , to be observable , must result from a two - body decay typically giving a lifetime @xmath170 sec , it is easy to see that the three - body decay lifetime ( from @xmath171 ) is shorter than the age of the universe , unless @xmath172 the limit coming from the single neutrino channel gives therefore a strong enough limit on @xmath129 to render irrelevant the limits one could get from all other possible @xmath5 decay channel , except from the @xmath173 channel .
it is easy to see that the latter channel gives up to a very good approximation the stronger constraint , @xmath174 which holds in order to avoid overproduction of galactic center @xmath175 kev photon from overproduction of positrons .
the limit on the lifetime of a dm particle with mass below @xmath176 mev is given by @xmath177 @xcite .
as for the lower bound , it comes from the fact that , if one decreases too much @xmath129 , one gets a decay into a photon and a neutrino which is too slow to account for any observable photon line that could be detected in the future .
for instance , let us consider the x - ray line recently reported with energy @xmath1 kev and flux @xmath178 @xcite . assuming a standard dm density along the line of sight , such a line , if better confirmed experimentally , could be understood from a dm decay into a photon and a neutrino , if the lifetime is @xmath179 s @xcite .
it is interesting to stress that such kind of lines could in principle be accounted for by any model leading to one of the fermion radiative operator reported in ref .
@xcite ( for the non - millicharged case ) or by any of the millicharged operators considered here , provided the operator can match the constraint that @xmath41 must be a lepton doublet and @xmath40 must be a singlet ( as its mass must lie in the kev range ) .
mass , also holds for lower mass . to explain this recently reported line
, one would not need necessarily to assume a fermion dm candidate .
the scalar or vector operators given in this reference , or in section [ scalarvector ] below , could also account for it , provided there exists another lighter scalar or vector particle to accompany the photon in the decay final state .
to distinguish among these operators appears to be hopeless , given the fact that the associated neutrino flux is basically unobservable at these energies .
for various other possible explanations of this line , see refs .
@xcite . ] for the millicharge option this latter requirement excludes the dimension - five operator but not the three dimension - six operators of eqs .
( [ o6fdpsisig])-([o6fd2sig ] ) with @xmath88 the scalar sm doublet . for these operators
one has nevertheless to check that for such low masses , and given the stringent constraints on the millicharge of neutrinos , one can get a lifetime of order the one needed .
the radiative lifetime one gets for any of these three operators is the same @xmath180 which , for the parameter values indicated , is about the one needed . given the uncertainties on the experimental flux needed , on the dm lifetime needed , and on the bounds on the neutrino millicharge from red giants , at the effective theory approach level one concludes that a small millicharge for the neutrino could be at the origin of this @xmath0-ray line , or more generally of observable kev - mev low energy lines , provided there is new physics around the corner at colliders and provided that the millicharge of the @xmath159 or @xmath160 is close to its upper bound . in other words , despite of the very stringent bounds which exist on them , neutrino millicharges could consequently have an observable effect in the form of a x - ray line .
note also that within the kev - mev mass range discussed here , for @xmath181 kev , the generation of the relic density for a fermionic sm singlet dm can nevertheless be challenging @xcite .
finally , remark that in the kinetic mixing scenario , imposing as above that @xmath182 , gives the constraint @xmath183 , which means @xmath184 .
after studying the possibility for a millicharged dm of the fermionic type to emit an observable @xmath0-ray line through its decay , we now turn to the scalar and vector dm cases .
if the dm particle is of the scalar or vector type , there also exist operators that could a priori lead to such a line . for what concerns the emission of cosmic rays , the phenomenology of these operators turns out to be similar to the fermion case .
therefore , in the following , we will limit ourselves to the determination of the operators and to a few additional general comments which slightly distinguish these scenarios from the fermion case .
due to angular momentum conservation considerations , a scalar dm particle can not decay to a scalar and a massless gauge boson .
therefore , it can only decay to a photon and another massive spin - one particle , which necessarily carries the same millicharge . as a result ,
the scalar case is similar to the vector case .
the operators that can allow a decay of a scalar dm particle into a vector particle and a photon , could also hold for a vector dm particle decaying into a scalar and a photon . in principle , a vector dm particle could nevertheless also decay in a different way , into a photon and another vector particle .
note that a necessary condition for the vector dm to acquire a millicharge is to be a complex field , therefore associated to a non - abelian gauge group . for the scalar - vector - photon case we found only one dimension - five operator @xmath185 and two dimension - six operators @xmath186 as said above , in these operators the dm particle can be either one of the scalar particle or a vector particle , the latter in one of the covariant derivative of eq .
( [ osv6fdd ] ) or in one of the non - abelian hidden sector @xmath187 field strengths of eqs ( [ osv5])-([osv6ff ] ) .
the photon can show up from one of the @xmath187 field strengths , through gauge boson mixing .
this occurs for instance if , on top of a kinetic mixing between the hypercharge gauge boson and a @xmath6 gauge boson , there is , through symmetry breaking , a mixing between this @xmath6 gauge boson and gauge boson(s ) of the new non - abelian symmetry ( whose field strength is @xmath187 ) . in this way
the @xmath187 field strength provides a photon field proportionally to the millicharge of the complex gauge boson in this field strength . ) and a @xmath188 @xmath187 field strengths , both field strengths must be taken in their quintuplet combination , which means that the scalar field is a quintuplet . ]
note that no operator containing only covariant derivatives remains .
this can be shown using equations of motion and rotating away non - canonical kinetic terms .
operator ( [ osv6fdd ] ) is equivalent to op .
( [ osv6ff ] ) up to operators that do not produce monochromatic photons . for the two operators of eqs .
( [ osv5])-([osv6ff ] ) , the emission of a @xmath5 is always suppressed by the millicharge squared , even if the scalars are non - sm singlet .
this stems from the fact that the @xmath5 as the @xmath0 can come only from the field strength in these operators , not from a covariant derivative as for the fermion operators .
this means that one gets the eq .
( [ ratio_a ] ) prediction even if the scalars are not sm singlets ( up to corrections in @xmath189 for @xmath190 ) .
of course , as for the fermion case , one can saturate the ratio of eq .
( [ ratio_a ] ) only if there is no cosmic ray production from @xmath7 decay ( in the stueckelberg case ) .
the latter could arise from the decay where the photon is replaced by a @xmath7 if this decay ( which is not suppressed by the value of the millicharge squared ) is not kinematically forbidden and if the @xmath7 subsequently decays into sm particles . as for the operator of eq .
( [ osv6fdd ] ) , it can lead to a non - millicharge suppressed production of @xmath5 and @xmath83 from the covariant derivatives , if the scalars are non - sm singlets .
its phenomenology is therefore very similar to the one of the fermion operators .
finally , about the possibility that a vector dm particle could decay into a photon and another vector particle , one could think about operators with only @xmath187 field strengths , for instance @xmath191 , through a similar mechanism where a @xmath187 could give a photon through gauge boson mixing .
we did not find any simple realization of such a possibility .
in summary , there are very few ways of probing the dm hypothesis that can really be considered in a systematic and model - independent way . however , for the decay of an absolutely neutral dm particle into a @xmath0-ray line , this turns out to be feasible @xcite .
this stems from the facts that , on the one hand , the use of an effective theory is fully justified , slow enough decay can naturally be explained from a much higher scale physics , and , on the other hand , it turns out that there are very few operator structures of this kind .
@xcite considered the usual scenario where the dm particle is absolutely neutral so that the photon appears in the operator through a field strength ( i.e. typically from a charged particle in a loop ) . here
we show that , for the same reasons , such a study can also be systematically carried out in the less considered scenario where the dm is millicharged , having therefore a tree - level coupling to the photon through a covariant derivative , either from an ad - hoc millicharge , or through mixing of the @xmath29 gauge boson with another @xmath6 gauge boson . to the emission of a @xmath0-ray line from such operators
is associated the emission of a continuum of cosmic rays .
the monochromatic photon to cosmic ray flux ratio is determined by the sm quantum numbers of the field on which the covariant derivative applies ( and in one case also crucially on the dm mass ) , and if this particle is not a sm singlet on the value of its millicharge .
this leads to upper bounds on the intensity of the @xmath0-ray line produced , given in fig .
[ fig : kin_mix ] .
this figure shows that if the dm is only charged under the dark sector , it can lead to a line matching the present experimental sensitivities without overshooting the bound on the flux of antiprotons and diffuse photons . on the contrary , when the particle emitting the photon from its millicharge is also charged under the sm , the cosmic rays constraints are much stronger than direct searches for spectral lines .
therefore , in this case , if a line were to be detected with energy above the @xmath5 mass and with about the present experimental sensitivity , it could not be explained in such a way .
such a conclusion can also hold for @xmath129 far below the @xmath5 mass . for the massless hidden gauge boson case ( and the massive case where @xmath192 is both below the gev scale and smaller than @xmath129 ) relevant additional constraints show up imposing that the two - body decay width to a @xmath147 ( z ) leads to a lifetime longer than the age of the universe . combining this constraint with the direct detection bounds on a millicharge ,
an observable @xmath0-ray line requires small values of the dark charge @xmath141 . as for a decay into a neutrino and a photon ,
given the stringent constraints that exist on the millicharge of a neutrino , the @xmath5 mediated decay into three neutrinos , or into a neutrino and a electron - positron pair , this possibility is forbidden unless @xmath129 is below the mev scale . for lower masses , and down to the kev scale , an observable line induced in this way
is not excluded by these considerations .
such a neutrino millicharge scenario could even be at the origin of the recently reported , yet to be confirmed , 3.5 kev x - ray line .
we acknowledge stimulating discussions with m. gustafsson and m. tytgat .
this work is supported by the fnrs - frs , the iisn , the fria and the belgian science policy , iap vi-11 .
thanks the ifae - uabarcelona theory group for its hospitality .
in the case in which @xmath193 but is not a singlet , the denominator of eq .
( [ ratio_ddpsidpsi ] ) takes the value @xmath194 in the case in which @xmath195 , the denominator takes the value @xmath196 l. bergstrom and h. snellman , phys .
d * 37 * ( 1988 ) 3737 ; s. rudaz , phys .
d * 39 * ( 1989 ) 3549 ; a. bouquet , p. salati and j. silk , phys . rev .
d * 40 * ( 1989 ) 3168 .
m. doro _ et .
[ cta consortium ] , astroparticle physics * 43 * ( 2013 ) 189 - 214 .
[ arxiv:1208.5356 [ astro-ph.im ] ] .
m. ackerman _ et .
al . _ , [ fermi - lat collaboration ] , phys .
d * 88 * ( 2013 ) 082002 .
[ arxiv:1305.5597[astro - ph.he ] ] .
n. fornengo , p. panci and m. regis , phys .
d * 84 * ( 2011 ) 115002 [ arxiv:1108.4661 [ hep - ph ] ] .
l. chuzhoy and e. w. kolb , jcap * 0907 * ( 2009 ) 014 [ arxiv:0809.0436 [ astro - ph ] ] .
o. adriani _ et al .
_ [ pamela collaboration ] , phys . rev . lett .
* 105 * ( 2010 ) 121101 .
[ arxiv:1007.0821 [ astro-ph.he ] ] .
n. f. bell , a. j. galea and k. petraki , phys .
d * 82 * ( 2010 ) 023514 [ arxiv:1004.1008 [ astro-ph.he ] ] .
m. frigerio , t. hambye and e. masso , phys .
x * 1 * ( 2011 ) 021026 [ arxiv:1107.4564 [ hep - ph ] ] .
e. bulbul , m. markevitch , a. foster , r. k. smith , m. loewenstein and s. w. randall , [ arxiv:1402.2301 [ astro-ph.co ] ] .
a. boyarsky , o. ruchayskiy , d. iakubovskyi and j. franse , [ arxiv:1402.4119 [ astro-ph.co ] ] . h. ishida , k. s. jeong and f. takahashi , [ arxiv:1402.5837 [ hep - ph ] ] .
d. p. finkbeiner and n. weiner , [ arxiv:1402.6671 [ hep - ph ] ] .
t. higaki , k. s. jeong and f. takahashi , [ arxiv:1402.6965 [ hep - ph ] ] .
j. jaeckel , j. redondo and a. ringwald , [ arxiv:1402.7335 [ hep - ph ] ] . | if a @xmath0-ray line is observed in the near future , it will be important to determine what kind of dark matter ( dm ) particle could be at its origin .
we investigate the possibility that the @xmath0-ray line would be induced by a slow dm particle decay associated to the fact that the dm particle would not be absolutely neutral .
a `` millicharge '' for the dm particle can be induced in various ways , in particular from a kinetic mixing interaction or through the stueckelberg mechanism .
we show that such a scenario could lead in specific cases to an observable @xmath0-ray line .
this possibility can be considered in a systematic model - independent way , by writing down the corresponding effective theory .
this allows for a multi - channel analysis , giving in particular upper bounds on the intensity of the associated @xmath0-ray line from cosmic rays emission .
our analysis includes the possibility that in the two - body decay the photon is accompanied with a neutrino .
we show that , given the stringent constraints which hold on the millicharge of the neutrinos , this is not an option , except if the dm particle mass lies in the very light kev - mev range , allowing for a possibility of explanation of the recently claimed , yet to be confirmed , @xmath1 kev x - ray line . |
heavy quarks @xmath3 play an important role in many searches for physics beyond the standard model ( sm ) with tau ( @xmath4 ) leptons .
i will focus here on the possibility of new physics effects associated to the scalar sector for which a connection between heavy quarks and the tau lepton is established due to their large mass .
the discovery of a sm - like higgs boson by the atlas and cms collaborations seems to confirm that a scalar sector is responsible for the breaking of the ew gauge symmetry @xcite .
the strong hierarchy of the fermion mass spectrum implies that the sm higgs couplings to light quarks and leptons are very suppressed .
the heaviest fermions offer then a unique place where exotic effects associated to the higgs sector can be sizable due to their larger masses .
i will discuss two topics within this context : * lepton flavour violating ( lfv ) higgs decays and higgs mediated lfv @xmath4 decays . * violation of lepton universality in @xmath5 decays due to charged scalar contributions at tree level .
let us assume that the recently discovered higgs boson around @xmath6 gev has lepton flavour violating couplings .
while such couplings are strongly constrained in the case of light leptons due to @xmath7 and @xmath8 to @xmath9 conversion in nuclei searches , couplings of the type @xmath10 and @xmath11 receive much weaker constraints from the current limits on lfv @xmath4 decays @xcite . in refs .
@xcite it was estimated that the lhc should be able to set stronger bounds on the hypothetical lfv higgs couplings with the amount of data accumulated at the moment .
the interactions we are interested are described by @xmath12 the chirality projectors @xmath13 are denoted as usual , the parameters @xmath14 are taken as free complex quantities to be bounded experimentally .
here we consider only @xmath10 lfv couplings but most of the arguments presented will hold also for @xmath11 .
the cms collaboration has recently performed a search for lfv decays of the @xmath6 gev higgs boson @xcite , setting the upper bound on the branching ratio ( br ) @xmath15 at 95% , the possibility to explain this excess within the framework of the type iii 2hdm has been discussed in ref .
@xcite . ] the partial decay width for this process is given by @xmath16 assuming that the total decay width of the higgs boson is simply the sm higgs total width plus @xmath17 one can extract the limit @xmath18 @xcite .
the top quark plays a very important role in the search for lfv higgs decays .
gluon fusion , together with vector boson fusion and associated higgs production with a vector boson , constitute the relevant higgs production mechanisms in the search for @xmath19 decays . the sm higgs coupling to gluons is described in the heavy top mass limit by the effective lagrangian @xmath20 note that the dimension five operator @xmath21 comes suppressed by the ew scale @xmath22 gev instead of the top mass , showing a non - decoupling behavior . assuming that all the diagonal higgs couplings are equal to their sm value one can translate the cms limits on @xmath23 into the following upper bounds : @xmath24 within the scenario considered , the current cms limit on @xmath25 puts the three - body lfv @xmath4 decays beyond the reach of belle ii . _
( a ) : higgs production via gluon fusion and later decay into a pair @xmath26 with @xmath27 .
( b ) contribution to the lfv decay @xmath28 due to lfv scalar interactions .
_ , title="fig:",width=124,height=64 ] _ ( a ) : higgs production via gluon fusion and later decay into a pair @xmath26 with @xmath27 .
( b ) contribution to the lfv decay @xmath28 due to lfv scalar interactions .
_ , title="fig:",width=124,height=90 ] ( a ) ( b ) heavy quarks are crucial for the estimation of the @xmath29 , @xmath30 and @xmath31 decay rates considered in eq . . the same non - decoupling behavior of heavy quarks in eq
is also relevant at very low energies when describing higgs mediated @xmath8 to @xmath9 conversion in nuclei @xcite and lfv semileptonic @xmath4 decays @xcite , see fig .
[ fig : introii ] . _
( a ) scalar mediated one - loop contribution to @xmath32 .
( b ) two - loop contribution of the barr - zee type to @xmath32._,title="fig:",width=124,height=79 ] _ ( a ) scalar mediated one - loop contribution to @xmath32 .
( b ) two - loop contribution of the barr - zee type to @xmath32._,title="fig:",width=124,height=86 ] + ( a ) ( b ) _ constraints on lfv higgs couplings from @xmath33 searches at b - factories and lfv higgs decays at the lhc .
we assume that the @xmath6 gev higgs has lfv couplings and take all diagonal higgs couplings to be sm - like as a benchmark .
figure adapted from ref .
_ , width=298,height=222 ] the lfv radiative decay @xmath34 is described by the effective lagrangian , @xmath35 here @xmath36 represents the high energy scale where the new physics appears .
the dimension five operators in eq .
appear at the loop level in renormalizable theories .
dipole transitions require a chirality flip between the initial fermion state and the final one .
scalar mediated contributions to @xmath34 at the one - loop level involve three chirality flips , two in the yukawa couplings and one in the fermion propagator , see fig .
[ fig : intro ] ( a ) .
this accidental suppression is avoided at the two - loop level in diagrams of the barr - zee type @xcite . in particular
, the wilson coefficients @xmath37 receive large contributions from two - loop diagrams with intermediate heavy quarks as the one shown in fig .
[ fig : intro ] ( b ) .
similar contributions also enter in the three - body decays @xmath38 and @xmath33 .
the complementarity between low energy searches for lfv via @xmath33 decays and searches for lfv higgs decays is summarized in fig .
[ fig : gluonic ] .
the babar collaboration has reported an excess with respect to the standard model ( sm ) in exclusive semileptonic transitions of the type @xmath39 @xcite .
more specifically , they have measured the ratios @xmath40 normalized by the corresponding light lepton modes @xmath41 .
these measurements present an excess of @xmath42 ( @xmath43 ) and @xmath44 ( @xmath45 ) with respect to the sm @xcite .
this excess would imply a violation of lepton universality at the @xmath46 level .
the most natural explanation to this excess would be a new charged mediator entering at tree - level which couples preferentially to the tau lepton .
_ charged scalar contribution to the exclusive semileptonic decay @xmath47 ( pink - dashed propagator)._,width=166,height=128 ] multi - higgs doublet models predict the existence of charged scalars which would contribute at tree level in @xmath47 decays , see fig . . moreover , by imposing the natural flavour conservation condition ( nfc ) to avoid dangerous flavour changing neutral currents one obtains that the charged scalar couplings to fermions are proportional to the fermion masses .
i will consider here the aligned two - higgs - doublet model ( a2hdm ) @xcite ; charged higgs couplings are still proportional to the fermion masses in this framework and all the versions of the 2hdm with nfc are recovered in particular limits .
charged higgs interactions with fermions are parametrized in the a2hdm by @xmath48 d\ , \nonumber \\ & + \
, \varsigma_l\ , \bar{\nu } m_l \mathcal p_r l \bigr\ } \ ; + \;\mathrm{h.c . } \ , .\end{aligned}\ ] ] the parameters @xmath49 ( @xmath50 ) are complex family universal parameters that need to be determined by experimental data .
the matrix @xmath51 represent the quark mixing matrix and @xmath52 ( @xmath50 ) are the diagonal fermion mass matrices . _
@xmath53 cl allowed regions in the plane @xmath54 considering different observables which receive a charged higgs contribution at tree level._,width=275,height=238 ] considering leptonic @xmath2 and @xmath55 meson decays in which the charged higgs also enters at tree level , it is not possible to accommodate the current excess in @xmath56 within the a2hdm @xcite .
[ fig : tr ] shows the allowed regions by the different observables considered in the @xmath54 plane .
it can be observed that the preferred values by @xmath57 are in conflict with the allowed region by @xmath55-meson leptonic decays . as a by - product of this result , we know that all versions of the 2hdm with nfc ( this includes the type ii 2hdm ) can not accommodate the excess in @xmath56 .
interestingly , all the observables can be accommodated within the a2hdm if @xmath45 is not considered in the fit .
the reason for this is that @xmath43 is sensitive to the scalar combination of charged higgs couplings while leptonic @xmath2 and @xmath55 decays are sensitive to the pseudoscalar combination .
due to the vectorial nature of the @xmath58 , @xmath45 is sensitive to the pseudoscalar combination which is already constrained by leptonic meson decays .
if the current excess in exclusive semileptonic @xmath39 transitions persists , it will be crucial to measure differential distributions in these decays .
differential distributions in the momentum transfer and polarization fractions of the @xmath58 meson can be very useful to disentangle different kinds of new physics , see refs .
@xcite and references therein .
heavy quarks @xmath3 play a crucial role in many circumstances when searching for new physics using @xmath4 leptons .
i have focused on new physics effects associated to the scalar sector .
due to their large masses compared with that of the other fermions , heavy quarks and the @xmath4 lepton are expected to have stronger couplings with scalar particles from an extended higgs sector .
i discussed the role of heavy quarks when probing for @xmath26 ( @xmath59 ) lepton flavour violation associated to the scalar sector via higgs or @xmath4 decays .
the non - decoupling behaviour of the top - quark in the effective @xmath60 vertex is crucial for the search of @xmath61 ( @xmath41 ) at the lhc .
heavy quarks play also a special role in the description of lfv @xmath4 decays because they enter in two - loop diagrams of the barr - zee type as the one shown in fig .
[ fig : intro ] ( b ) . the one - loop scalar mediated contribution shown in fig . [ fig : intro ] ( a ) is accidentally suppressed so that two - loop diagrams can dominate over the one - loop contribution .
i have also discussed possible violations of lepton universality in @xmath47 decays , taking into account the observed excess by the babar collaboration in these modes .
the current excess in @xmath62 can not be accommodated within the framework of the a2hdm @xcite . as a by - product of this analysis one
obtains that none of the two - higgs - doublet models with natural flavour conservation ( this includes the type ii 2hdm ) can explain this excess either .
recent theoretical developments have reduced the discrepancy between theory and experiment in @xmath43 to about @xmath63 @xcite .
the situation for @xmath45 is still unclear .
new belle measurements of @xmath43 and @xmath45 will be useful to clarify the current situation .
future measurements of differential distributions in @xmath47 decays would provide a crucial handle to discriminate between different new physics alternatives in case the current excess in @xmath56 persists .
_ et al . _
[ atlas collaboration ] , _ `` observation of a new particle in the search for the standard model higgs boson with the atlas detector at the lhc , '' _ phys .
b * 716 * ( 2012 ) 1 [ arxiv:1207.7214 [ hep - ex ] ] .
s. chatrchyan _ et al . _
[ cms collaboration ] , _ `` observation of a new boson at a mass of 125 gev with the cms experiment at the lhc , '' _ phys .
b * 716 * ( 2012 ) 30 [ arxiv:1207.7235 [ hep - ex ] ] .
g. blankenburg , j. ellis and g. isidori , _
`` flavour - changing decays of a 125 gev higgs - like particle , '' _ phys . lett .
b * 712 * ( 2012 ) 386 [ arxiv:1202.5704 [ hep - ph ] ] .
r. harnik , j. kopp and j. zupan , _
`` flavor violating higgs decays , '' _ jhep * 1303 * ( 2013 ) 026 [ arxiv:1209.1397 [ hep - ph ] ] .
talk by jan steggemann on behalf of the cms collaboration , _ searches for higgs bosons decaying to lepton pairs with the cms detector " _ , 4 of july 2014 .
international conference of high energy physics ( ichep ) .
a. crivellin , m. hoferichter and m. procura , _
`` improved predictions for @xmath65 conversion in nuclei and higgs - induced lepton flavor violation , '' _ phys .
d * 89 * ( 2014 ) 093024 [ arxiv:1404.7134 [ hep - ph ] ] .
a. celis , v. cirigliano and e. passemar , _
`` lepton flavor violation in the higgs sector and the role of hadronic @xmath4-lepton decays , '' _ phys .
d * 89 * ( 2014 ) 1 , 013008 [ arxiv:1309.3564 [ hep - ph ] ] . j. d. bjorken and s. weinberg , _
`` a mechanism for nonconservation of muon number , '' _ phys .
rev . lett .
* 38 * ( 1977 ) 622 . s. m. barr and a. zee ,
_ `` electric dipole moment of the electron and of the neutron , '' _ phys .
* 65 * ( 1990 ) 21 [ erratum - ibid .
* 65 * ( 1990 ) 2920 ] .
j. p. lees _ et al .
_ [ babar collaboration ] , _
`` evidence for an excess of @xmath66 decays , '' _ phys .
* 109 * ( 2012 ) 101802 [ arxiv:1205.5442 [ hep - ex ] ] .
j. f. kamenik and f. mescia , _
`` @xmath67 branching ratios : opportunity for lattice qcd and hadron colliders , '' _ phys .
d * 78 * ( 2008 ) 014003 [ arxiv:0802.3790 [ hep - ph ] ] .
a. datta , m. duraisamy and d. ghosh , _ `` diagnosing new physics in @xmath71 decays in the light of the recent babar result , '' _ phys .
d * 86 * ( 2012 ) 034027 .
p. biancofiore , p. colangelo and f. de fazio , _
`` on the anomalous enhancement observed in @xmath72 decays , '' _ phys .
d * 87 * ( 2013 ) 7 , 074010 .
j. a. bailey , a. bazavov , c. bernard , c. m. bouchard , c. detar , d. du , a. x. el - khadra and j. foley _ et al .
_ , _ `` refining new - physics searches in @xmath73 decay with lattice qcd , '' _ phys .
* 109 * ( 2012 ) 071802 [ arxiv:1206.4992 [ hep - ph ] ] .
d. becirevic , n. kosnik and a. tayduganov , _ `` @xmath74 vs. @xmath75 , '' _ phys . lett .
b * 716 * ( 2012 ) 208 [ arxiv:1206.4977 [ hep - ph ] ] . | in this talk i discuss the role of heavy quarks in new physics searches with tau leptons .
i focus on new physics effects associated to the scalar sector which are naturally enhanced for the heaviest fermions due to the large hierarchy of the fermion masses .
i will discuss two topics within this context : lepton flavour violation in the @xmath0 ( @xmath1 ) sector and violations of lepton universality in tauonic @xmath2 decays .
tau lepton , lepton flavour violation , lepton universality |
as new and deeper galaxy redshift surveys are being completed , a definite picture of the local galaxy distribution is slowly emerging .
this distribution is characterized by large voids surrounded by sheets of clustered matter ( see e.g. el - ad et al . 1996a , 1997 ) , and can be well quantified by statistical descriptors like the power spectrum ( see e.g. park et al .
1994 , tadros & efstathiou 1995 , tadros & efstathiou 1996 , lin et al .
1996 ) . in a few years , surveys like the sloan digital sky survey ( loveday 1996 ) and the two degree field redshift survey ( colless 1998 , maddox 1998 ) will extend our knowledge of the clustering of the luminous matter by at least a factor of ten both in depth and coverage
. however , mapping the luminous galaxies is clearly not enough to fully understand the history and geography of our universe .
first , even the deepest surveys so far planned do not reach beyond @xmath7 .
secondly , we already know that a prominent component of the matter , perhaps a dominant one , is not bright enough to be included in the present surveys .
ranging from truly dark matter , to dwarf galaxies , to very low surface brightness galaxies , to lyman-@xmath1 clouds , the systems which can easily escape detection are many and varied .
their importance in solving crucial questions , such as whether the voids traced by bright galaxies are really empty of matter ( szomoru et al .
1994 , stocke et al . 1995 ) , which are the paths of galaxy formation , what is the role of galaxy interactions , can not be underestimated . in this respect ,
the large number of absorption lines seen in quasar spectra , the so called ly@xmath1 forest , is an important tracer of the intervening matter distribution along the qso lines of sight at any redshift in the observable universe ( @xmath8 ) .
even if the ly@xmath1 forest is a one - dimensional distribution , the fact that a line of sight can contain at high redshift ( @xmath9 ) up to a few hundreds of absorption lines makes the statistics significant .
while galaxies are associated with emitting objects , the ly@xmath1 clouds represent the gas detected through the absorption of hi and are not necessarily associated with stars . in this sense
they are matter tracers different from any other .
what is the fraction of baryonic matter in the igm is still matter of controversy .
hydrodynamical simulations of the ly@xmath10 forest ( rauch & haehnelt 1995 , miralda escud et al . 1996 ,
rauch et al .
1997 , weinberg et al .
1997 , bi & davidsen 1997 ) indicate that this fraction is large for redshift @xmath11 ( @xmath12 ) and collapsed objects in the young universe represent probably a small correction .
however observations of the ly@xmath1 forest with column density in the range @xmath13 by kim et al .
1997 using keck / hires qso spectra give @xmath14 in @xmath15 .
the discrepancy is probably due to the fact that in the simulations the ( photo and collisional ) ionisation of the igm is assumed to be higher than that assumed by kim et al . , who neglected collisional ionisation because they derived an upper limit to the temperature of the gas @xmath16 k from the measured doppler parameter of the ly@xmath17 lines .
the study of signatures in the distribution of ly@xmath1 lines has been performed mainly using the traditional tool for the galaxy distribution : the two point correlation function ( tpcf ) . while it has been clear for many years ( sargent et al .
1980 ) that low resolution spectroscopy was unable to give a definitive answer to the clustering of the ly@xmath1 forest for scales of @xmath18 km s@xmath19 ( no signal was found for larger scales ) , the advent of high resolution spectroscopy in the last few years has started a more controversial discussion on the presence of clustering at small scales , and on the evolution of the same with redshift .
a weak signal has been found on small scales ( @xmath20 km s@xmath19 , webb 1987 , rauch et al .
1992 , chernomordik 1995 , cristiani et al .
1995 ) . more recently , a detection of redshift evolution of the tpcf , being stronger at lower redshift , in a large sample of high column density ly@xmath1 lines ( @xmath21 ) ( cristiani et al .
1997 , savaglio et al .
1999 ) has been only marginally confirmed by a similar study by kim et al .
( 1997 ) , keeping open the question whether the ly@xmath1 forest does cluster and at which scales .
this controversy leads to the natural consequence that it is necessary to find other tools of investigation for the same problem since the tpcf is neither the only statistical technique nor necessarily the most appropriate .
pando & fang ( 1996 ) have shown that the scaling properties of the ly@xmath1 forest can be studied using the wavelet decomposition analysis .
the results are that clustering is present up to large scales ( 20 @xmath22 mpc ) and that the 1d power spectrum is significantly different from a poissonian distribution .
counting technique ( carbone & savaglio 1996 , savaglio & carbone 1997 ) has the advantage of avoiding boundary effects , and has given a positive signal evolving with redshift ( up to about 3 @xmath22 mpc at @xmath23 and up to about 20 @xmath22 mpc at @xmath24 ) .
the ly@xmath17 forest has been studied also by hui et al .
( 1997 ) , by deriving the expected column density distribution given a theoretical power spectrum
. we will investigate the distribution of the lyman-@xmath1 clouds by two different , but related , descriptors : the 1-dim power spectrum , and the density of neighbours , or conditional density .
the former is particularly suited to detecting preferential scales in the distributions , since it decomposes the system in plane waves . here
`` preferential scale '' means simply a wavelength whose fourier coefficient is significatively different than expected assuming a gaussian distribution of the fourier coefficients .
a preferential scale of clustering has sometimes been claimed in the galaxy distribution ( broadhust et al .
1991 ; einasto et al .
1997 ) , so that it is interesting to see if something similar holds for the lyman-@xmath17 clouds .
the 1-dim power spectrum is however intrinsically a very noisy quantity , since the fourier coefficients are statistically independent quantities in the limit of a gaussian field .
the main advantage is that , as a consequence of the statistical independence , their statistics is straightforward . on the other hand ,
the conditional density is an integral quantity , because it counts the average density of clouds within a distance @xmath25 of another cloud ; its signal is stable , but its statistics can only be approximated via montecarlo simulations .
the conditional density has a reduced scatter with respect to the tpcf , and is therefore easier to compare to theoretical models , although the information on specific scales is spread over a large interval .
once the analysis of the dataset is completed , we can compare it to theoretical models of the density fluctuation field .
the primary goal is to derive the lyman-@xmath1 power spectrum in the linear regime as a function _ both of redshift and column density_. we will therefore derive a biasing function @xmath26 , defined as the ratio of the cloud spectrum to a theoretical cdm spectrum at scales large enough to be in the linear regime of gravitational clustering . since we will consider clouds at @xmath27 , we can assume safely that comoving scales larger than 3 mpc / h are linear , i.e. the variance in spheres of 3 mpc / h at @xmath27 is less than unity .
although we will report data even at lower scales , all the relevant numerical results are obtained using only scales larger than 3 mpc / h . for simplicity ,
we normalize the cdm spectrum to the value that matches the present day cluster abundance , i.e. @xmath28 ( white , efstathiou & frenk 1993 ) ; a different choice of @xmath29 , for instance normalizing to the microwave background fluctuations or to the present galaxy clustering , simply rescale @xmath30 by a constant factor . since at the high redshifts of our data the space - time geometry has an important effect , we test the dependence of our results considering in the following two values of the cosmological density : the inflationary value @xmath31 and the observationally preferred @xmath32 .
the final product is the cloud power spectrum approximated as @xmath33 where @xmath34 is a cdm spectrum normalized to @xmath35 , evolving with @xmath36 as @xmath37 , where @xmath38 is the fluctuation growth function in the linear regime .
we adopt the approximation ( padmanabhan 1993 ) @xmath39 which is valid for @xmath40 and @xmath41 ( the effect of a finite @xmath42 is actually negligible on @xmath43 ) . for @xmath31
this gives @xmath44 as expected .
the biasing function @xmath45 expresses therefore the evolution of the clustering of the clouds both versus redshift and column density .
one can disentangle the evolutionary path in @xmath36 and @xmath46 only by means of further assumptions .
although this task is left to future work , we obtain the preliminary result that the ly@xmath47 clustering evolves linearly only if at the same time the average column density decreases strongly with time .
our method is based on the parametric estimation of the power spectrum of fitted absorption lines .
a completely different method has been proposed in a recent work by croft et al .
they can successfully recover both the shape and the amplitude of the linear power spectrum of initial mass fluctuations ( assumed to be gaussian ) from the one dimensional power spectrum derived from qso lyman@xmath17 spectra , artificially created by realistic hydrodynamic cosmological simulations .
it is premature to compare the results of the two approaches , also because croft et al .
apply the method to only one observed qso line of sight ; the results on the spectrum of q1244 + 231 at @xmath48 are consistent with a standard cdm model with normalization @xmath49 .
this conclusion applies also to our findings , although we find a very marked dependence on redshift and on column density . [ cols="<,<,^,^,^,^,^,^,^,^,^ " , ] [ f1 ] let @xmath50 be the redshift of the @xmath51-th lyman-@xmath1 cloud along a line of sight , and let @xmath52 be its mass .
we will later assume a unitary mass for the clouds .
the fourier transform of the cloud distribution is then @xmath53\exp ( ikr)dr\ ] ] where @xmath54\ ] ] is the cloud distance , @xmath55 is the total length of the cloud system , @xmath56 is the mass density , @xmath57 its average .
we can put @xmath58 , where @xmath59 is a dirac delta function .
it follows that the total mass along the line of sight is @xmath60 , and that @xmath61 , where @xmath62 is the average mass .
then @xmath63 where @xmath64 is the window transform .
the window integral is approximated in @xmath65 equally spaced bins , which should be taken as small as possible , down to the resolution limit of the data .
finally , the power spectrum is defined as @xmath66 by this definition , we have @xmath67 .
if the clouds are uncorrelated , then on average @xmath68 for @xmath69 ; therefore , we get @xmath70 and then we obtain the pure shot - noise power spectrum @xmath71 in the limit of an infinite sample ( i.e. for @xmath72 ) @xmath73 we derived the equations including the mass , but in the following we put @xmath74 ( from which @xmath75 ) because the relation between the observed hi column density and the mass is dominated by poorly known parameters like the cloud size and shape . as a simple test , we arbitrarily assumed @xmath76 and found no qualitative difference .
we will evaluate @xmath77 at @xmath78 , so as to obtain @xmath79 coefficients @xmath80 .
the fourier coefficients @xmath81 , being linear combinations of many independent random variables @xmath82 , tend to be distributed as gaussian variables .
the power spectrum coefficients , sum of the square of two gaussian variables , are distributed as a @xmath83 distribution with two degrees of freedom , i.e. as an exponential @xmath84 if the central limit theorem conditions are not fulfilled , e.g. the @xmath85-modes are correlated or the data are strongly non - gaussian , the distribution of @xmath86 deviates from the exponential . to estimate the noise level @xmath71 for a given line of sight , we generate one thousand simulated distributions of lines , distributing randomly the lines .
then , we estimate @xmath71 by fitting the distribution of power spectrum coefficients with the exponential ( [ pdf ] ) , and average over the realizations .
the agreement with the expected value ( [ expnoi ] ) is excellent . from eq .
( [ pdf ] ) we see that the variance of @xmath71 is @xmath87 however , this gives only an idea of the spread of the spectrum coefficients around the noise . to be more quantitative ,
we compare the cumulative distribution function ( cdf ) of the spectrum coefficients @xmath88 with the cdf of the simulated distribution .
the noise cdf is @xmath89 therefore , a peak of height @xmath90 among @xmath91 peaks is statistically significant if @xmath92 is very low , e.g. @xmath93 .
= 6.5 cm [ f3 ] = 8.9 cm = 8.9 cm = 8.9 cm the one - dimensional power spectrum is plagued by the aliasing problem stressed by kaiser & peacock ( 1991 ) : spurious high signal - to - noise peaks may be induced by the 1-dim geometry .
however , since we compare the ly@xmath17 spectrum to random realizations with the same geometry , and not to a theoretical significance level , our method is able to assess the probability that random realizations produce a given spectrum peak .
the second quantity we use is the normalized density of neighbours , i.e. @xmath94 where the sum extends to all the lines within a distance @xmath25 of a given line assumed as a center , and the average extends over all the centers , and where @xmath95 is the expected poisson density . at the scales at which a clustering signal is detected , @xmath96 evaluating @xmath97 the following problem arises .
let us order the lyman-@xmath17 clouds along a line of sight on the @xmath98 axis , with increasing distance @xmath98 .
consider the lyman-@xmath17 clouds within a distance @xmath25 from the @xmath51-th cloud at position @xmath99 : there are clouds on the right , at @xmath100 , and clouds on the left , at @xmath101 .
suppose now the left set , for instance , is cut by the boundary of the sample .
we would count the clouds contained in it in @xmath102 even if , in fact , the left set contains lines only up to a distance smaller than @xmath25 .
if the clustering decreases with scale , as expected , the left set cut by the boundary would have on average a higher @xmath102 than if it were not cut .
in other words , including sets cut by the boundary amounts to averaging different scales in @xmath102 . this would result in a systematic boundary effect , an effect which increases for large radii and for large clustering . to avoid this , we include only the clouds in sets which are not cut by the boundary : that is , around each cloud , we take only the right set if the left one is cut , and vice versa .
let us write down here the relations between the two quantities introduced , the power spectrum and the density of neighbours , and the relation between these and the 3-dim power spectrum @xmath103 .
we have @xmath104 @xmath105 where @xmath106 is the 1-dim top - hat window function , @xmath107 .
the theoretical spectra are in real space .
as is well known , the peculiar velocities suppress the power of the real space spectrum at small scales and enhance it at large scales .
the redshift space spectrum @xmath108 is then a function of the real space spectrum @xmath109 of the factor @xmath110 , and of the cloud velocity dispersion @xmath111 along the line of sight .
the overall effect can be modeled according to the following formula derived by peacock & dodds ( 1994 ) : @xmath112- \nonumber \\ & & \frac{\exp ( -y^2)}{4y^4}[\beta ^2(3 + 2y ^2)+4\beta y^2 ] \label{dist}\end{aligned}\ ] ] where @xmath113 .
we will investigate scales from 1 to 100 mpc/@xmath114 the lower limit of the scale is close to the typical transverse size of the ly@xmath1 clouds , being of the order of a few hundred @xmath22 kpc with no strong redshift evolution ( crotts et al . 1997 , smette et al . 1995 ) . at scales below 1 or 2
mpc/@xmath115 it is very difficult to model accurately the lyman-@xmath17 power spectrum , due to various effects : the line - of - sight velocity dispersion , the thermal broadening of the lines , the procedure of line fitting , the non - linear clustering .
croft et al .
( 1998 ) found that a correction factor @xmath116 with @xmath117=1.5 mpc/@xmath115 may account empirically for all these small scale effects .
although we included some of the effects above mentioned to correct the spectrum at small scales , we work out the numerical results using only scales larger than 3 mpc/@xmath115 , where the corrections are small . we take for the velocity dispersion @xmath118 : simulations give in fact peculiar velocities about 100 km s@xmath19 or less ( miralda escud , 1996 , rauch et al .
1997 , bi & davidsen 1997 ) , while observations of quasar pairs give velocity differences between common absorption lines less than about 100 km s@xmath19 ( smette et al . , 1995 , dinshaw et al . , 1994 , @xmath24 )
although the level of peculiar velocity is expected to vary with redshift , we fix its value to 100 km / sec both because the range of @xmath36 we investigate is not very large , and because the correction is anyway important only at scales less than 1 mpc/@xmath115 .
let us mention that the expected igm spectrum deviates from the cdm spectrum at small scales also because of the clustering suppression below the igm jeans mass .
this deviation can be accounted for by a term which depends essentially on the igm temperature and redshift ( bi & davidsen 1997 ) .
as long as @xmath16k ( kim et al .
( 1997 ) found that at @xmath119 less than 19% of the clouds with @xmath120 can have @xmath121k ) , the scale at which a sensible deviation occurs is smaller than 0.5 mpc @xmath22 , for all relevant @xmath36 . in the following we therefore neglect this factor .
we calculated the ps for all our cloud systems , and averaged over the high and low redshift subsets mentioned in section 2 . in this section
we assume for simplicity @xmath31 , since here we will not compare the results to a cdm model , but rather to poisson distributions .
we have tested that putting @xmath32 in the distance - redshift relation does not modify the conclusions . in fig .
[ f1 ] we show the power - to - noise ratios @xmath122 for the average of the six higher and , separately , of the seven lower redshift qso lines of sight in the range from 1 to 100 mpc @xmath22with cut - offs @xmath12313.3 and @xmath12413.8 ( the wavelength being @xmath125 ) .
since we have expressed our spectra in units of the noise , we can take the average straightforwardly : the difference in line density due to evolution and to possible other observational effects is automatically taken into account .
the result is that there is no evidence of deviation from the poisson spectra , i.e. that there are no preferential scales , or periodicity
, in the clustering .
although there are a few isolated peaks above 2@xmath126 , they are not stronger and more frequent than one would get in a pure poisson distribution .
this is quantitatively reported in fig .
[ f2 ] , where we display the cumulative distribution of the peaks in all the 13 spectra at two column density cut - off : the cdf is always less than 1@xmath126 from poisson , even if we can have signals as strong as 9.5 times the noise ( top left panel of fig . [ f2 ] ) .
the same holds true for all the other cuts in column density .
we conclude that the power spectrum analysis of the distribution of lines does not detect clustering on any particular scale .
this does not necessarily mean , of course , that there is no clustering above poisson ; a suitable filtering of the data could in fact reveal a significant signal .
however , a filtering in @xmath85-space has not in general a direct geometrical meaning .
it is therefore simpler to consider an integral statistics in real space , such as the integral density of neighbours , and to compare this statistics with the theoretical expectation .
as one can see from the results of the previous section , the 1-dim power spectrum is extremely noisy , and can not easily be compared with the theoretical predictions . to perform this comparison ,
we adopt then in this section the integral density of neighbours .
the theoretical linear cdm - like galaxy spectrum in real space can be written as @xmath127 where @xmath128 is the transfer function of davis et al .
( 1985 ) , @xmath129 is a theoretical free parameter ( equal to @xmath130 in cdm models if @xmath85 is in @xmath115 mpc@xmath19 ) and @xmath131 is the normalization factor .
we normalize putting the present variance in 8 mpc @xmath22 spherical top - hat cells equal to the cluster abundance value @xmath132 ( white , efstathiou & frenk 1993 ) .
we corrected the spectrum for the non - linear enhancement as suggested by peacock & dodds ( 1996 ) , but we found that at scales larger than 3 mpc/@xmath115 the effect is negligible .
as anticipated , we will compare quantitatively our results to the data only at scales larger than 3 mpc/@xmath115 .
we have to determine the expected one - dimensional spectrum from the 3d spectrum in redshift space , @xmath133 and to evolve it back .
this gives the spectrum @xmath134 putting @xmath135 in eq .
( [ p1rho ] ) we can obtain the expected density of neighbours . in fig .
[ f3 ] we report the theoretical density of neighbours for some interesting cases .
naturally , the expression ( [ back ] ) is an oversimplification .
both the non - linear and the redshift corrections induce @xmath36-dependent distortions on the spectrum . an accurate modeling of these distortions requires extensive @xmath136-body calculations .
however , these distortions are important only at small scales , where in any case the uncertainties in the estimation of the lyman-@xmath17 are a major hurdle to a precise comparison .
[ f5 ] according to the standard model of biased gravitational instability , and to the findings of cristiani et al .
( 1996 ) , we expect that the clustering increases with decreasing redshift , and with increasing column density . however to fully understand how the 1d distribution of hi column density in qso spectra represents
the true 3d distribution of hydrogen in the universe at different redshifts is not an easy task .
given an hi column density associated with an absorption line , the real mass of the ly@xmath1 clouds depends on the ionisation correction @xmath137 in the cloud which gives the total hydrogen column density , and on the geometry of the cloud . both these quantities can be function of redshift due to the evolution of the uv background flux @xmath138 which photoionises the cloud and to the evolution of the gravitational instabilities in the clouds . to estimate a pure redshift evolution of the 3d power spectrum of the intergalactic hi , it is thus necessary to consider any redshift evolution of the ly@xmath17 clouds not associated with gravitational clustering .
since this is not a trivial problem , in this paper we do not investigate the evolution of the clustering in redshift at a fixed value of the column density , but , as a preliminary task , we map the variation of clustering both with redshift and column density . we leave to future work the comparison of this evolution map with the theory or with @xmath136-body and hydrodynamic simulations , a necessary step in the reconstruction of the history of the igm .
we show in figs .
[ f4 ] the density of neighbours for several of our subsets .
the error bands are poisson errors estimated via montecarlo simulations ; we adopted the poisson errors since the level of cosmic variance is always negligible .
for the set of the six highest redshift systems ( @xmath139 ) there is no detectable clustering , no matter what the cut in the column density is .
for the lowest redshift subset ( @xmath140 ) , we see a clear increase in clustering with @xmath141 . in fig .
( [ fig4omega ] ) the same low redshift subset is shown , now assuming an open universe with @xmath32 . comparing with the cdm curve
, we see that the lines with @xmath142 are near or above the cdm linear predictions , both for the flat and the open geometry . we also compared the results for the two line sets 1 and 1@xmath143 listed in table [ t1 ] ; the results are compatible within the errors .
it is interesting to observe that the cdm model is not an adequate fit for the low @xmath36 , high column density data , independently of the bias factor .
the normalized density of neighbours is in fact always significantly less than unity at large scales ( that is , there is large scale anticorrelation ) contrary to the cdm model .
we could not reproduce accurately this large scale feature even by varying the shape parameter @xmath144 . in fig .
[ f5 ] we show the clustering trend in the six subsets of table [ t2 ] , for @xmath145 , assuming @xmath31 .
the clustering increases regularly from @xmath146 to @xmath147 , although the increase is considerably faster at low @xmath36 . assuming as a reference the cdm predictions with @xmath148 and linear evolution
, we can estimate from the sets in table [ t2 ] the biasing function @xmath149 which expresses for any given average @xmath150 and any given redshift , the relation between the real clustering and the linear cdm predicted clustering ( the average @xmath150 has been obtained excluding the very few lines with @xmath151 ) .
clearly , @xmath152 means that the data are more clustered than the linear - extrapolated cdm distribution ; @xmath153 implies an under - clustering of the data . in fig .
[ f6 ] we plot the function @xmath154 .
as already mentioned , the real evolution is likely to be both in column density and clustering strength .
if one assumes arbitrarily that the average column density does not change with time , then the evolution in redshift turns out to be extremely fast .
in flat space , clustering growth as fast as @xmath155 or faster seem to fit the trend we obtain for high column density cut - offs .
one can obtain more reasonable values , i.e. values of closer to the linear trend _ only _ by assuming at the same time a decrease of the average column density with decreasing redshift . to give a rough approximation , mostly valid at redshift near 2
, we can fit the function @xmath156 as @xmath157 where @xmath158=3 , and @xmath159 atoms/@xmath160 if @xmath31 , and as @xmath161 where @xmath158 is the same , and @xmath162 atoms/@xmath160 for @xmath32 . the relative error in the exponents
is in all cases of the order of 30% .
the bias function shows quantitatively the clustering trend both in column density and redshift . as an interesting consequence
, we see that the clustering evolves linearly , i.e. @xmath156 is constant in redshift , only if @xmath163 ( or as @xmath164 for the open case ) .
notice that this conclusion is independent of the overall amplitude of the bias function ; it depends only on the relative level of clustering of the ly@xmath17 distributions .
two caveats have to be kept in mind .
first , our assumption of a scale - independent bias may prove incorrect .
however , we used it as a simple way to parameterize the lyman-@xmath17 spectrum in terms of the cdm spectrum , and as such it looks , _ a posteriori _ , as an acceptable approximation .
second , the clustering evolution of the clouds need not be linear , even at the scales we are investigating .
in fact , along with the linear evolution due to the gravitational instability , the clustering may evolve also in response to the evolution of the ionising sources .
if , as shown e.g. by bagla ( 1998 ) , the evolution of the clustering of the first collapsed objects is far from linear at @xmath165 , then the lyman-@xmath17 clouds evolution could also be non - linear .
the aim of this paper was to reconstruct the power spectrum of the lyman-@xmath1 clouds as a function both of redshift and average column density , and to compare it to theoretical cdm models .
we first estimated directly the line - of - sight power spectrum : we concluded that there is no evidence of characteristic scales in the cloud distribution , contrary to what is found in some galaxy surveys .
since the direct reconstruction of the power spectrum is too noisy to derive useful comparisons with theoretical expectations , we employed next the integrated density of neighbours .
we assumed that the cloud spectrum can be fitted by @xmath166 where @xmath167 characterizes the lyman-@xmath17 line bias with respect to a cdm model normalized to the cluster abundance .
this fit performs acceptably well for scales between 1 and 10 mpc@xmath168 , while the data are systematically anticorrelated on larger scales .
we found that the ly@xmath1 clouds are clustered with power similar to the theoretical cdm only at redshift near 2 and average column density higher than 10@xmath169 hi atoms/@xmath160 , being underclustered ( @xmath170 ) in all the other cases .
a reasonable expectation is that the cloud evolution is linear for most of the scales considered here . imposing this condition upon the fits ( [ fit ] ) and ( [ fit04 ] )
we conclude that the average column density must decrease in time as @xmath171 ( or as @xmath172 for a @xmath32 universe ) . for the popular model with a cosmological constant such that @xmath173 the results of the open case applies to a good approximation , since @xmath174 does not affect significantly the clustering evolution .
notice that the function @xmath175 depends on assuming the specific parametric form ( [ parform ] ) and in particular on the clustering redshift evolution of the dark matter spectrum .
however , the trend in @xmath36 we find is so fast that it seems difficult to account for it without a strong redshift decrease of @xmath176 .
it is clear that the dataset investigated here is too small to give definitive results .
one of the problems is that the most recent and higher cut - off samples are also the most sparse , and thus give higher statistical uncertainty .
one conclusion is however rather firm : the cloud clustering can be explained only by assuming a simultaneous increase in correlation and decrease in average column density with redshift .
carbone v. , savaglio s. , 1996 , mnras , 282 , 868 carswell r. f. , lanzetta k. m. , parnell h. c. , webb j. k. , 1991 , apj , 371 , 36 chernomordik v.v . , 1995 , apj , 440 , 431 colless m. m. , 1998 , phil .
lond.a , in press , astro - ph/9804079 cristiani s. , dodorico s. , fontana a. , giallongo e. , savaglio s. , 1995 , mnras , 273 , 1016 cristiani s. , dodorico s. , dodorico v. , fontana a. , giallongo e. , savaglio s. , 1997 , mnras , 285 , 209 croft r. , weinberg d. , katz , n. , & hernquist l. , 1998 , apj , 495 , 44 crotts a. , fang y. , 1998 , apj , 502 , 16 davis m. , efstathiou g. , frenk c. , white s.d.m . , 1985 , apj , 292 , 371 dinshaw n. , impey c. d. , foltz c. b. , weymann r. j. , chafee f. h. , 1994 , apj , 437 , l87 dodorico v. , cristiani s. , dodorico s. , fontana a. , giallongo e. , a&as , 127 , 217 el - ad h. , piran t. & da costa l.n . , 1996 , apj , 462 , l13 el - ad h. , piran t. & da costa l.n . , 1997 , mnras , 287 , 790 hu e. m. , kim t. , cowie l. l. , songaila a. , 1995 , aj , 110 , 1526 hui l. gnedin n. y. , zhang y. , 1997 , apj , 486 , 599 khare p. , srianand d. g. , york d. g. , green r. , welty d. , huang k.l . ,
bechtold j. , 1997 , mnras , 285 , 167 kim t.s . , hu e. m. , cowie l. l. , songaila a. , 1997 , aj , 114 , 1 kulkarni v. p. , huang k. , green r. f. , bechtold j. , welty d. e. , york d. g. , 1996 , mnras , 279 , 197 lin h. , kirshner r. p. , shectman s. a. , landy s. d. , oemler a. , tucker d. l. , schechter p. l. , 1996 , apj , 471 , 617 loveday j. , 1996 , proc .
xxxist rencontres de moriond , astro - ph/9605028 lu l. , sargent w. l. w. , barlow t. a. , 1997 , apj , 484 , 131 maddox s. j. , 1998 , in ` large scale structure : tracks and traces ' , world scientific , in press miralda escud j. , cen r. , ostriker j. p. , rauch m. , 1996 , apj , 471 , 582 padmanabhan t. , _ structure formation in the universe _ , cambridge university press , 1993 .
pando j. , fang l.z .
, 1996 , apj , 459 , 1 park c. , vogeley m.s .
, geller m.j . , & huchra j.p . , 1994 ,
apj , 431 , 569 peacock j. a. & dodds s.j . , 1994 , mnras , 267 , 1020 peacock j. a. & dodds s.j . , 1996 , mnras , 280 , 19 rauch m. , carswell r.,f . , chafee f. h. , foltz c. b. , webb j. k. , weymann r. j. , bechtold j. , green r. f. , 1992 , apj , 390 , 387 rauch m. , haehnelt m. , 1995 , mnras , 275 , l76 rauch m. , haehnelt m. , steinmetz m. , 1997 , apj , 481 , 601 rodrguez - pascual p.m. , de la fuente a. , sanz j.l .
, recondo m.c .
, clavel j. , santos - lle m. , wamsteker w. , 1995 , apj , 448 , 575 sargent w. l. w. , young p. j. , boksenberg a. , tytler d. , 1980 , apjs , 42 , 41 savaglio s. , carbone v. , 1997 , proc .
sesto workshop on _ dark and visible matter in galaxies _ , eds .
m. persic and p. salucci , a.s.p .
conference series , vol .
117 , p491 savaglio s. , cristiani , dodorico s. , fontana a. , giallongo , e. , molaro p. , 1997
, a&a , 318 , 347 savaglio s. , ferguson h. c. , brown t. m. , et al .
, 1999 , apj , l5 smette a. , robertson j.g . ,
shaver p.a .
, reimers d. , wisotzki l. , khler th
. , 1995 , a&as , 113 , 199 stocke j.t . ,
shull m. , penton s. , donahue m. , carilli c. , 1995 apj , 451,24 szomoru a. , guhathakurta p. , van gorkom j.h .
, knapen j.h . , weinberg , d.h .
, & fruchter a.s . , 1994 ,
aj , 108 , 491 tadros h. & efstathiou g. , 1995 , mnras , 282 , 1381 tadros h. & efstathiou g. , 1996 , mnras , 285 , l5 webb j. k. , 1987 , in hewett a. , burbidge g. , fang l. z. eds . ,
124 , observational cosmology .
reidel , dordrecht , p. 803
weinberg d. h. , miralda
escud j. , hernquist l. , katz n. , 1997 , apj , 490 , 564 white s. d. m. , efstathiou g. p. , & frenk c. s. , 1993 , mnras , 262 , 1023 | we investigate the clustering properties of 13 qso lines of sight in flat space , with average redshifts from @xmath0 to 4 .
we estimate the 1d power spectrum and the integral density of neighbours , and discuss their variation with respect to redshift and column density . we compare the results with standard cdm models , and estimate the power spectrum of lyman-@xmath1 clustering as a function both of redshift and column density .
we find that @xmath2 ) there is no significant periodicity or characteristic scale ; @xmath3 ) the clustering depends both on column density and redshift ; @xmath4 ) the clustering increases linearly only if at the same time the hi column density decreases strongly with redshift .
the results remain qualitatively the same assuming an open cosmological model .
2@xmath5 @xmath6 hsd epsf.tex |
in recent years , the experimental realization of landau - zener - stckelberg ( lzs ) interferometry@xcite in several systems has emerged as a tool to study quantum coherence under strong driving .
lzs interferometry is realized in two - level systems ( tls ) which are driven with a time periodic force .
the periodic sweeps through an avoided crossing in the energy level spectrum result in successive landau - zener transitions .
the accumulated phase between these repeated tunneling events gives place to constructive or destructive interferences , depending on the driving amplitude and the detuning from the avoided crossing .
these lzs interferences have been observed in a variety of quantum systems , such as rydberg atoms,@xcite superconducting qubits,@xcite ultracold molecular gases,@xcite quantum dot devices,@xcite single spins in nitrogen vacancy centers in diamond,@xcite nanomechanical circuits,@xcite and ultracold atoms in accelerated optical lattices.@xcite several other related experimental and theoretical works have studied lzs interferometry in systems under different shapes of periodic driving,@xcite in two coupled qubits,@xcite in optomechanical systems,@xcite and the effect of a geometric phase.@xcite furthermore , experiments in superconducting flux qubits under strong driving have allowed to extend lzs interferometry beyond two levels.@xcite in this later case , the multi - level structure of the flux qubit , with several different avoided crossings in the energy spectrum , exhibited a series of diamond - like interference patterns as a function of dc flux detuning and microwave amplitude.@xcite driven two - level systems have been extensively studied theoretically in the past . under strong time - periodic driving fields , phenomena such as coherent destruction of tunneling ( cdt ) @xcite and multiphoton resonances @xcite
have been analyzed .
the influence of the environment has been studied within the driven spin - boson model,@xcite applying various techniques like the path - integral formalism,@xcite or the solution of the time dependent equations for the populations of the density matrix,@xcite either as an integro - differential kinetic equation,@xcite or considering for weak coupling the underlying bloch - redfield equations,@xcite or using the decomposition of the quantum master equation into floquet states,@xcite or performing a rotating - wave approximation.@xcite however , the theoretical results for driven tls in contact with a quantum bath @xcite present some discrepancies with the experiments of lzs interferometry , particularly in the flux qubit.@xcite in fact , these works typically predict population inversion,@xcite which has not been observed in the experiments of ref . . in a recent work
we showed,@xcite that for the case of a non - resonant detuning , population inversion arises for very long driving times and it is mediated by a slow mechanism of interactions with the bath .
this long - time asymptotic regime was not reached in the experiment . here , we will study in full detail the lzs interference patterns of the flux qubit , by considering the dependence with dc flux detuning in addition to the dependence with microwave amplitude . to this end , we calculate numerically the finite time and the asymptotic stationary population of the qubit states using the floquet - markov approach,@xcite solving a realistic model of the flux qubit in contact with an ohmic quantum bath . for the time scale of the experiments we find a very good agreement with the diamond patterns of ref .. for longer time scales ,
we find a dynamic transition within the first diamond ( which corresponds to the two level regime ) , manifested by a symmetry change in the structure of the lzs interference pattern .
we also consider the case of an structured quantum bath , which can be due to a squid detector with a resonant plasma frequency @xmath0 .
different types of lzs interference patterns can arise in this case , depending on the magnitude of @xmath0 .
the paper is organized as follows . in sec.ii
we present the model for the flux qubit and we review the basics of lzs interferometry . in sec .
iii , we show our results for the emergence of a dynamic transition in the lzs interference pattern at long times . in sec . iv we give our conclusions and an appendix
is included to describe in detail the floquet - markov formalism used in this paper .
superconducting circuits with mesoscopic josephson junctions are used as quantum bits@xcite and can behave as artificial atoms.@xcite .
a well studied circuit is the flux qubit ( fq ) @xcite which , for millikelvin temperatures , exhibits quantized energies levels that are sensitive to an external magnetic field .
the fq consists on a superconducting ring with three josephson junctions@xcite enclosing a magnetic flux @xmath1 , with @xmath2 .
the circuits that are used for the fq have two of the junctions with the same josephson coupling energy , @xmath3 , and capacitance , @xmath4 , while the third junction has smaller coupling @xmath5 and capacitance @xmath6 , with @xmath7 .
the junctions have gauge invariant phase differences defined as @xmath8 , @xmath9 and @xmath10 , respectively . typically the circuit inductance can be neglected and the phase difference of the third junction is : @xmath11 .
therefore , the system can be described with two independent dynamical variables .
a convenient choice is @xmath12 ( longitudinal phase ) @xmath13 ( transverse phase ) . in terms of this variables , the hamiltonian of the fq ( in units of @xmath14 ) is:@xcite @xmath15 with @xmath16 and @xmath17 .
the kinetic term of the hamiltonian corresponds to the electrostatic energy of the system and the potential term corresponds to the josephson energy of the junctions , given by @xmath18 typical flux qubit experiments have values of @xmath19 in the range @xmath20 and @xmath21 in the range @xmath22.@xcite in quantum computation implementations @xcite the fq is operated at magnetic fields near the half - flux quantum , @xmath23 , with @xmath24 . for @xmath25 ,
the potential @xmath26 has two minima at @xmath27 separated by a maximum at @xmath28 .
each minima corresponds to macroscopic persistent currents of opposite sign , and for @xmath29 ( @xmath30 ) a ground state @xmath31 ( @xmath32 ) with positive ( negative ) loop current is favored .
, for the qubit parameters @xmath33 and @xmath34 considered throughout this work .
energy is measured in units of @xmath14 and flux in units of @xmath35 .
the gaps at the avoided crossings are indicated as @xmath36 .
black lines correspond to `` longitudinal '' modes and red lines to `` transverse '' modes ( see text for description ) .
the blue arrows at the bottom of the plot illustrate the relation between ac driving amplitudes and level crossing positions for a particular dc flux detunning @xmath37 .
the indicated values correspond to the edges of the spectroscopic diamonds of fig.[las ] for the given @xmath37 . ] in fig.[fig - espec ] we plot the lowest energy levels @xmath38 as a function of @xmath39 , obtained by numerical diagonalization of @xmath40.@xcite in this case we set @xmath41 and @xmath42 , close to the experimental values employed in flux qubits experiments.@xcite negative ( positive ) slopes in fig.[fig - espec ] , correspond to eigenstates with positive ( negative ) loop current .
a gap @xmath36 opens at the avoided crossings of the @xmath43-th level of positive slope with the @xmath44-th level of negative slope at @xmath45 . for our choice of device parameters ,
the avoided crossing of the two lowest levels at @xmath46 has a gap @xmath47 ( in units of @xmath14 ) . at larger @xmath48
, we find the avoided crossing with a third level at @xmath49 with gap @xmath50 , and next the avoided crossing at @xmath51 with gap @xmath52 .
( there are also energy levels that correspond to excited transverse modes,@xcite plotted with red lines in fig.[fig - espec ] , but they have a negligible contribution to the dynamics considered here ) .
the two - level regime , involving only the lowest eigenstates at @xmath53 and @xmath54 , corresponds to @xmath55 , such that the avoided crossings with the third energy level are not reached . in this case , the hamiltonian of eq .
( [ ham - sys ] ) can be reduced to a two - level system@xcite @xmath56 in the basis defined by the persistent current states @xmath57 , with @xmath58 and @xmath59 the ground and excited states at @xmath60 .
the parameters of @xmath61 are the gap @xmath62 and the detuning energy @xmath63 . here
@xmath64 is the magnitude of the loop current , which for our case with @xmath42 and @xmath33 is @xmath65 ( in units of @xmath66 ) .
landau - zener - stckelberg ( lzs ) interferometry is performed by applying an harmonic field on top of the static field with @xmath67 if the ac driving amplitude is such that @xmath68 , which is fulfilled for @xmath69 , the quantum dynamics can be described within the two - level regime . in this case
, we use @xmath70 in the hamiltonian of eq.([htls ] ) , with @xmath71 and @xmath72 . when @xmath73 the central avoided crossing at @xmath60 is reached within the range of the driving amplitude , @xmath74 . in this case the periodically repeated landau - zener transitions at @xmath60 give place to lzs interference patterns as a function of @xmath75 and @xmath76 , which are characterized by _
multiphoton resonances_@xcite and _ coherent destruction of tunneling_,@xcite as we describe below .
there are _ multiphoton resonances _
when@xcite @xmath77 ( where @xmath78 is written in units of @xmath79 and energies in units of @xmath14 ) .
if @xmath80 , these resonances are at @xmath81.@xcite calling @xmath82 , the @xmath83-resonance condition can also be written as @xmath84 .
an example of the dynamic behavior in a @xmath85 resonance is shown in fig.[fo1](a ) , calculated as described in ref .
( see also the appendix for details of the calculation ) . in this case
the fq is driven with frequency @xmath86 ( for @xmath87ghz it corresponds to @xmath88mhz ) .
the dc detuning is @xmath89 , corresponding to @xmath90 ( since @xmath91 ) .
the fq is started at @xmath92 in the ground state @xmath93 , and the probability of having a positive loop current is calculated , @xmath94 . since for @xmath95 we have @xmath96 , the initial probability is @xmath97 .
the red - dashed line in fig [ fo1](a ) shows the time dependence of @xmath98 .
the resonant dynamics is clearly seen : the fq oscillates coherently between positive and negative current states .
therefore , the probability @xmath98 oscillates between @xmath99 and @xmath100 , having a time averaged value of @xmath101 .
in contrast , fig.[fo1](b ) shows the time dependence of @xmath98 in an off - resonant case , for @xmath102 .
we see that for off - resonance the fq stays in the positive loop current state , since @xmath98 fluctuates around @xmath103 .
state as a function of time , for a flux qubit driven with frequency @xmath86 ( in units of @xmath79 ) .
( a ) @xmath104 ( second resonance ) and @xmath105 .
( b ) @xmath106 ( off resonance ) and @xmath107 .
( c ) same as ( b ) but for a long time scale .
red dashed line shows the results in the non dissipative case and the black line corresponds to the dissipative case with an ohmic bath at @xmath108 . ] for @xmath109 , it has been shown , in a rotating wave approximation , that the average probability @xmath110 near a @xmath111 photon resonance is@xcite @xmath112 at the resonance , @xmath113 , eq.([l2ls ] ) gives @xmath114 and away of the resonance @xmath115 .
the width of the resonance is @xmath116 , with @xmath117 the bessel function of the first kind .
this gives a quasiperiodic dependence as a function of @xmath76 for @xmath75 fixed near the resonance .
in particular , at the zeros of @xmath117 , the resonance is destroyed , giving @xmath118 instead of @xmath101 , a phenomenon known as _ coherent destruction of tunneling_.@xcite plots of @xmath119 as a function of flux detuning @xmath75 and ac amplitude @xmath76 give the typical lzs interference patterns , which have been measured experimentally by oliver et al in flux qubits,@xcite and have also been observed in other systems . @xcite the eq.([l2ls ] )
corresponds to ideally isolated flux qubits , neglecting the coupling of the qubit with the environment .
the dynamics of the fq as an open quantum system is usually characterized by the energy relaxation time @xmath120 and the decoherence time @xmath121 .
several phenomenological approaches have taken into account relaxation and decoherence in lzs interferometry , obtaining a broadening of the lorentzian - shape @xmath111 photon resonances of eq.([l2ls ] ) .
for example , in ref .
a bloch equation approach is used , obtaining for zero temperature : @xmath122 with @xmath123 and @xmath124 .
similar results were obtained by berns et al,@xcite considering a pauli rate equation with an effective transition rate that adds to @xmath125 a driving induced rate @xmath126 .
the latter was obtained assuming that decoherence is due to a classical white noise in the magnetic flux , an approach that is valid for time scales smaller than the relaxation time and larger than the decoherence time , for @xmath127 .
the case of low frequency @xmath128 noise has been considered in a similar approach,@xcite finding a gaussian line shape for the @xmath83-resonances instead of the lorentzian line shape .
the description of the lzs interference patterns with phenomenological approaches like eq.([l2lsr ] ) gives a good agreement with most of the experimental results.@xcite however , they rely on approximations valid either for large frequencies , or for low ac amplitudes , or for time scales smaller than the relaxation time . here
we will study lzs interferometry using the floquet formalism for time - periodic hamiltonians,@xcite which allows for an exact treatment of driving forces of arbitrary strength and frequency .
the time - dependent schrdinger equation is transformed to an equivalent eigenvalue equation for the floquet states @xmath129 and quasi - energies @xmath130 , which can be solved numerically .
( see the appendix for details ) . to describe relaxation and decoherence processes , we consider that the qubit is weakly coupled to a bath of harmonic oscillators .
the markov approximation of the bath correlations is performed after writing the reduced density matrix @xmath131 of the qubit in the floquet basis , @xmath132.@xcite by following this procedure , the quantum master equation obtained for @xmath133 is valid for periodic driving forces of arbitrary strength .
( in contrast , the standard born - markov approach is valid for small driving forces ) .
the obtained floquet - markov quantum master equation is @xcite : @xmath134 the coefficients @xmath135 depend on the spectral density @xmath136 of the bath , the temperature @xmath137 , and on the qubit - bath coupling . using the numerically obtained floquet states @xmath129
, we can calculate the coefficients @xmath135 , and then we compute the solution of @xmath138 as described in the appendix . here , we will consider the dynamics of a fq coupled to a bath with an ohmic spectral density@xcite @xmath139 with @xmath140 a cutoff frequency .
the ohmic bath mimics an unstructured electromagnetic environment that in the classical limit leads to white noise .
we consider @xmath141 , corresponding to weak dissipation,@xcite and a large cutoff @xmath142 .
the bath temperature is taken as @xmath143 ( @xmath144 for @xmath145 ) . experimentally , the probability of having a state of positive or negative persistent current in the flux qubit is measured.@xcite the probability of a positive loop current measurement can be calculated in general as@xcite @xmath146\ ] ] with @xmath147 the operator that projects wave functions on the @xmath148 subspace , as described in the appendix . in fig [ fo1 ]
the black lines show the population @xmath149 as a function of time obtained from the numerical solution of eq.([drho ] ) , taking the ground state @xmath58 as initial condition , and compared with the isolated fq ( red dashed lines ) discussed in the previous section . in fig
[ fo1](a ) , for the @xmath150 resonance , we see that the population @xmath98 has damped oscillations that tend to the asymptotic average value of @xmath101 for large times . on the other hand , for the off - resonant case of fig [ fo1](b ) and ( c ) we see that there is a clear difference between the short time behavior of @xmath98 , shown in fig [ fo1](b ) , and the large time behavior shown in fig [ fo1](c ) .
moreover , we observe that @xmath98 tends to asymptotic values that are very different than in the isolated system . for the particular case shown in the plot
, we find population inversion , i.e @xmath151 , in the asymptotic long time limit , a phenomenon we discussed in ref . for off - resonant cases . in the following
, we will see how this change of behavior in the long time is reflected in the full lzs interference pattern as a function of @xmath75 and@xmath76 . as a function of the driving amplitude @xmath76 and dc detuning @xmath75 for ( a ) @xmath152 , time scale of the experiments , and ( b ) asymptotic regime for @xmath153 . in ( a )
we mark the ac amplitudes defined in the text that correspond to the edges of the spectroscopic diamonds for a given @xmath37 .
( c ) @xmath149 for @xmath106 as a function of the driving amplitude @xmath76 , at @xmath152 ( blue dashed line ) and asymptotic average population @xmath154 ( black line ) .
the calculations were performed for @xmath155 and ohmic bath at @xmath156mk .
vertical lines separate diamond regimes d1 , d2 , d3 , described in the text . , title="fig : " ] as a function of the driving amplitude @xmath76 and dc detuning @xmath75 for ( a ) @xmath152 , time scale of the experiments , and ( b ) asymptotic regime for @xmath153 .
in ( a ) we mark the ac amplitudes defined in the text that correspond to the edges of the spectroscopic diamonds for a given @xmath37 .
( c ) @xmath149 for @xmath106 as a function of the driving amplitude @xmath76 , at @xmath152 ( blue dashed line ) and asymptotic average population @xmath154 ( black line ) .
the calculations were performed for @xmath155 and ohmic bath at @xmath156mk .
vertical lines separate diamond regimes d1 , d2 , d3 , described in the text . , title="fig : " ] as a function of the driving amplitude @xmath76 and dc detuning @xmath75 for ( a ) @xmath152 , time scale of the experiments , and ( b ) asymptotic regime for @xmath153 .
in ( a ) we mark the ac amplitudes defined in the text that correspond to the edges of the spectroscopic diamonds for a given @xmath37 .
( c ) @xmath149 for @xmath106 as a function of the driving amplitude @xmath76 , at @xmath152 ( blue dashed line ) and asymptotic average population @xmath154 ( black line ) .
the calculations were performed for @xmath155 and ohmic bath at @xmath156mk .
vertical lines separate diamond regimes d1 , d2 , d3 , described in the text . , title="fig : " ] in figs.[las](a ) and [ las](b ) we show an intensity plot of @xmath149 as a function of @xmath75 and @xmath76 , calculated at a time near the experimental time scale of refs . , @xmath157 , with @xmath158 the period of the ac driving .
for @xmath69 we observe an lzs pattern modulated by multiphoton resonances and coherence destruction of tunneling , which can be described within a good approximation with eq.([l2lsr ] ) .
this lzs interference plot is very similar to the experimental results of ref .. furthermore , for higher ac amplitudes , we find a pattern of `` spectroscopic diamonds '' , also in good agreement with experiments.@xcite the diamond structure obtained for high ac amplitudes can be related to the energy level spectrum of fig.[fig - espec ] , for a fixed dc flux detuning @xmath159 , as follows.@xcite the first diamond , d1 , starts when the @xmath160 avoided crossing at @xmath161 is reached by the amplitude of the ac drive , _
i.e. _ when @xmath162 .
this defines an onset ac amplitude @xmath163 .
the first diamond ends when the nearest second avoided crossing is reached ( with gap @xmath164 ) at the ac amplitude @xmath165 .
then the second diamond , d2 , starts when the other second avoided crossing is reached , at @xmath166 , and it ends when the next avoided crossing is reached at @xmath167 .
similarly , the third diamond , d3 , starts at @xmath168 , and so on .
the analysis of the positions of the resonances as a function of @xmath76 and @xmath75 was the route followed in refs . to obtain the parameters characterizing the different avoided crossings of the flux qubit .
the fq of refs .
have short decoherence times ( @xmath169 ) and large relaxation times ( @xmath170 ) .
the typical duration of the driving pulses in these experimental measurements ( @xmath171 ) is in between this two time scales , @xmath172 . due to this time scale separation , a model with classical noise,@xcite valid for @xmath173 ,
can qualitatively explain the experimentally observed behavior of @xmath149 within the first diamond , through eq.([l2lsr]).@xcite moreover , a multilevel extension of the model of ref
. can also describe the higher order diamonds ( provided one gives as an input parameter the positions @xmath174 and the gaps @xmath36 of the avoided crossings).@xcite for large times @xmath175 one would expect naively , that after full relaxation with the environment , a blurred picture of the lzs interference pattern of figs.[las](a ) with broadened resonance lobes should be observed .
the asymptotic @xmath176 , averaged over one period @xmath177 , is shown in fig.[las](b ) .
[ @xmath178 can be calculated exactly after obtaining numerically the right eigenvector of @xmath179 with zero eigenvalue .
see the appendix for details .
] surprisingly , we see in fig.[las](b ) that the asymptotic behavior of @xmath149 gives a qualitatively different lzs interference pattern within the first diamond , and not a mere blurred version of fig.[las](a ) .
in particular , a cut at constant @xmath75 is shown in fig.[las](c ) . there
we see clearly that within the first diamond regime , @xmath180 , the value of @xmath149 at the experimental time scale is very different than the asymptotic value @xmath178 . on the other hand , beyond the first diamond , for @xmath181
the results of @xmath182 and @xmath178 are nearly coincident . as a function of the @xmath183 flux detuning @xmath75 for the flux qubit driven with @xmath107 and @xmath155 , and coupled to an ohmic bath at @xmath184mk ( for @xmath185 ) .
red line with squares : @xmath152 .
black line with circles : asymptotic ( @xmath153 ) average population @xmath154 .
blue line with triangles : time averaged population in the isolated system . horizontal orange line : indicates the @xmath186 level to help to identify when there is population inversion .
the flux detuning @xmath48 is normalized by @xmath187 , such that the @xmath83-photon resonances are at @xmath188 .
( b ) enlarged view around the @xmath85 resonance .
the green dashed line is a plot of the best fit with eq.([l2lsr ] ) , while a plot of eq.([l2ls ] ) ( not shown ) falls exactly over the blue dotted line . ] in fig.[fpf0 ] we show @xmath149 vs. @xmath75 , for @xmath107 , within the first diamond .
the blue line with triangles corresponds to the solution of the isolated system , which shows dips where @xmath189 that correspond to the @xmath83-photon resonances at @xmath190 .
the red line with squares corresponds to @xmath182 , which shows values of @xmath149 smaller than in the isolated case ( due to effect of decoherence and relaxation ) and in agreement with the experiments .
the asymptotic @xmath178 is also shown in fig.[fpf0 ] ( black line with circles ) , which is lower than @xmath182 .
fig.[fpf0](b ) shows in detail the behavior near the @xmath85 resonance .
the isolated case shows a dip with a lorentzian shape accurately described by eq.([l2ls ] ) .
the open system at the experimental time scale shows a broadened peak for @xmath182 , partially consistent with a description like eq.([l2lsr ] ) , except that near the resonance the behavior of @xmath149 vs @xmath75 becomes antisymmetric around @xmath190 . in the asymptotic steady regime , @xmath178
vs @xmath75 is antisymmetric in a wide region around @xmath190 , showing population inversion ( @xmath191 ) below the resonance , for @xmath192 .
+ the temporal evolution from symmetric to antisymmetric resonances can be seen in detail in fig.[figtemp ] , where we show an intensity plot of @xmath149 as a function of @xmath75 and the duration time @xmath193 , for @xmath107 .
two different temperatures are considered , a low temperature @xmath194 in figs.[figtemp](a ) and ( c ) , and @xmath195 in figs.[figtemp](b ) and ( d ) .
the relaxation time @xmath196 for the driving amplitude considered in fig.[figtemp ] is @xmath197 ( as we will see in sec.iiic , @xmath196 depends on @xmath76 ) .
figs.[figtemp](a ) and ( b ) are for duration times up to the time scale @xmath198 of the experiments , for the two different temperatures . we see that within this time scale @xmath199 , the resonances are symmetric . in figs.[figtemp](c ) and ( d ) the duration time is plotted in logarithmic scale and the evolution at very large times is shown .
we see that at a time @xmath200 the resonances start to change form , becoming asymmetric in the long time limit .
the asymmetry is stronger for low temperatures , in particular for @xmath201 there is full inversion of population on one side of the resonances . , to ( b ) triangular checkerboard pattern for asymptotic long times .
enlarged views of fig.[las](a ) and ( b ) , respectively , showing a sector of the first diamond including the @xmath202 multiphoton resonances .
the flux detuning @xmath48 is normalized by @xmath187 , such that the @xmath83-photon resonances are at @xmath188 .
, title="fig : " ] , to ( b ) triangular checkerboard pattern for asymptotic long times .
enlarged views of fig.[las](a ) and ( b ) , respectively , showing a sector of the first diamond including the @xmath202 multiphoton resonances .
the flux detuning @xmath48 is normalized by @xmath187 , such that the @xmath83-photon resonances are at @xmath188 .
, title="fig : " ] fig.[figtemp ] shows that , as a function of time , _ there is a dynamic transition manifested by a symmetry change in the structure of the lzs interference pattern_. the dynamic transition is from nearly symmetric resonances at short times ( @xmath173 ) to _ antisymmetric resonances _ at long times ( @xmath175 ) .
this is clearly illustrated in fig.[las2 ] where an enlarged view of a part of the first diamond is shown .
fig.[las2](a ) corresponds to @xmath182 and shows the characteristic features observed in experiments : ( i ) @xmath203 away from the resonances , ( ii ) there are resonance lobes where @xmath204 , and ( iii ) the @xmath83-resonance lobes are limited by the points where there is coherent destruction of tunneling ( cdt ) , given by @xmath205 . on the other hand , in the asymptotic regime , [ ( fig.[las2](b ) ] ,
the pattern of symmetric @xmath83-resonance lobes is replaced by a pattern of antisymmetric resonances , which form a triangular checkerboard picture defined by triangles with @xmath206 and @xmath207 alternatively , with their vertices located at the cdt points .
we name , in short , the former pattern of lzs interferometry as symmetric resonances " ( sr ) and the latter as antisymmetric resonances " ( ar ) .
that contribute to the relaxation rate . as a function of the driving amplitude @xmath76 .
the calculations were performed for @xmath106 , @xmath155 and an ohmic bath at @xmath208 ( for @xmath185 ) .
( b ) asymptotic average population @xmath154 for the same case .
the dashed vertical lines highlight one of the regions of @xmath76 where there is population inversion . ]
the asymptotic ar interference patterns show population inversion ( pi ) on one side of a multiphoton resonance , as seen in fig.[fpf0](b ) and figs.[figtemp](c ) and ( d ) .
this makes @xmath209 antisymmetric around the resonance .
if we fix @xmath75 at a value in between two resonances , the probability oscillates around @xmath210 as function of @xmath76 , showing pi whenever a `` triangle '' with @xmath211 is traversed .
this is shown in fig.[fo4](a ) .
the underlying mechanism of this population inversion can be understood by analyzing the contribution to relaxation of virtual photon exchange processes with the bath.@xcite as we show in the appendix , within the first diamond regime , the total relaxation rate @xmath212 can be decomposed as @xmath213 with @xmath214 the relaxation rates due to virtual @xmath83-photon transitions to bath oscillator states.@xcite in fig.[fo4](b ) we plot @xmath214 as a function of @xmath76 for the same @xmath75 considered in fig.[fo4](a ) .
we show the cases with @xmath215 , where @xmath216 describes the relaxation without exchange of virtual photons , corresponding to the `` conventional '' dc relaxation mechanism , while @xmath217 correspond to the the ac contribution due to the exchange of one virtual photon with energy @xmath218 .
we show in fig.[fo4 ] that , whenever there is population inversion , the dc relaxation terms vanish ( @xmath219 ) , while the @xmath220 term is the largest one .
this indicates that the relevant mechanism leading to pi is a transition to a virtual level at energy @xmath221 ( one photon absorption , @xmath222 ) , followed by a relaxation to the level @xmath223.@xcite and decoherence rates @xmath224 and @xmath225 as a function of the driving amplitude @xmath76 .
the calculations were performed for @xmath106 , @xmath155 and an ohmic bath at @xmath208 ( for @xmath185 ) .
vertical lines separate diamond regimes d1 and d2 described in the text . ]
to better understand why the ar patterns have not been observed yet in current experiments , we now analyze the time scales of relaxation and decoherence . to this end , we calculate numerically the full relaxation rate @xmath212 and the decoherence rates @xmath226 , from the eigenvalues of the @xmath179 matrix , as discussed in the appendix . in fig.[fo4c ] we show the relaxation rate @xmath212 and two decoherence rates @xmath226 , as a function of the driving amplitude @xmath76 for @xmath106 ( away from a @xmath83-resonance ) . for @xmath227
the relaxation rate @xmath212 corresponds to the @xmath228 measured experimentally , _
i.e. _ , @xmath229 when @xmath230 .
since to a good approximation the density matrix becomes diagonal in the floquet basis in the asymptotic regime,@xcite the decoherence rate @xmath224 shown in fig.[fo4c ] corresponds to the decoherence between the @xmath231 and the @xmath232 floquet states .
when @xmath230 , @xmath233 , since in this limit it corresponds to the decoherence between the two lowest energy levels .
we also plot in fig.[fo4c ] the rate @xmath225 , which is the decoherence rate between the @xmath234 and the @xmath235 floquet states .
at @xmath227 it corresponds to the decoherence between the third and the fourth energy level . for small @xmath76 , below the onset of the first diamond ( @xmath236 ) , the relaxation and decoherence rates stay in values similar to the undriven case .
however , when @xmath237 we see in fig.[fo4c ] that both rates depend strongly on @xmath76 . above the onset of the first diamond , the overall behavior is that the decoherence rate @xmath224 increases and the relaxation rate @xmath212 decreases as a function of @xmath76 .
therefore , within the first spectroscopic diamond , the difference between decoherence and relaxation is much larger than in the undriven case ( @xmath238 ) .
this explains the important difference between @xmath182 and @xmath178 , within the first diamond in fig.[las ] , due to the large time window where ar patterns can be observed for @xmath239 . beyond the first diamond , for @xmath181 , the relaxation rate @xmath212 increases strongly with @xmath76 , becoming nearly of the same order of the decoherence rates within the second diamond and above .
this behavior is a consequence of the fact that when more than two - levels are involved in the dynamics , there are several possible decay transitions between energy levels that contribute to a faster relaxation of the system .
therefore , in the second diamond and beyond decoherence and relaxation rates become comparable . for this reason ,
@xmath240 in this case , since the relaxation time is significantly reduced and thus @xmath241 .
although previous works have found pi for the asymptotic regime of two level systems,@xcite the time - dependent dynamics with different time scales has been overlooked .
in fact , the relevant point from our findings is that the asymptotic regime is difficult to reach in the experiment , since pi needs long times ( @xmath242 ) to emerge when mediated by the bath .
and the two virtual levels @xmath243 as a function of the flux detuning @xmath48 , for the flux qubit coupled with an structured bath with a resonant mode at frequency @xmath0
. the location of the `` virtual crossings '' at @xmath244 are also indicated .
the inset shows a plot of the spectral density @xmath136 of the structured bath . ] the measurement of the state of the fq is performed with a read - out dc squid , which is inductively coupled to the qubit.@xcite this modifies the bath spectral density by adding a resonant mode at the plasma frequency @xmath0 of the squid detector .
the so - called `` structured bath '' spectral density is given by@xcite @xmath245 with @xmath246 the width of the resonant peak at @xmath247 ( see the inset of fig.[majo ] for a schematic plot of @xmath136 ) .
as a function of the driving amplitude @xmath76 and dc detuning @xmath75 for ( a ) @xmath248 and ( b ) @xmath249 .
we indicate in the plot the regions corresponding to the regimes ( i ) , ( ii ) and ( iii ) described in the text .
, title="fig : " ] as a function of the driving amplitude @xmath76 and dc detuning @xmath75 for ( a ) @xmath248 and ( b ) @xmath249 .
we indicate in the plot the regions corresponding to the regimes ( i ) , ( ii ) and ( iii ) described in the text .
, title="fig : " ] here , we study the effect of this resonant mode at @xmath0 on lzs interferometry , in the case @xmath250 . to this end
, we calculate the asymptotic @xmath178 using the spectral density of eq.([jsb ] ) , considering different values of @xmath0 , with @xmath251 fixed . in fig.[lasp ]
we show the intensity plots of @xmath178 as a function of @xmath252 for @xmath248 and @xmath249 .
as it is evident , the diamond patterns are strongly affected by the structured bath . in the case of @xmath248 , we observe in fig.[lasp](a ) that the @xmath252 region formerly occupied by the first diamond in the ohmic case [ shown in fig.[las](b ) ] is now divided in three parts : two new sub - diamonds indicated as regimes ( i ) and ( iii ) , and the region in between them , indicated as regime ( ii ) . when lowering @xmath0 the regime ( iii ) becomes more predominant , as can be seen in fig.[lasp](b ) for @xmath249 . from now on , to describe the above mentioned changes of the first diamond , we restrict to the two lowest levels of the fq .
it has been shown that a two - level system coupled to an structured bath that has a localized mode at @xmath0 is equivalent to a two level system weakly coupled to a single mode quantum oscillator with frequency @xmath0 , and both coupled to an ohmic bath.@xcite in fact , most of the results discussed in this section can be qualitatively interpreted by assuming that at low energies there are two virtual levels at @xmath253 and @xmath254 , as sketched in fig.[majo ] .
these virtual levels are not stable and decay fast to their `` underlying '' energy level , _
i.e. _ , @xmath255 and @xmath256 .
the three regimes found in fig.[lasp ] can be understood by considering that there are two `` virtual crossings '' when @xmath257 , at the field detunings @xmath258 as shown schematically in fig.[majo ] .
the boundaries of these sub - diamonds can be defined in a similar way as in sec.iii.b , replacing @xmath259 by @xmath260 in the argument ( when @xmath261 ) .
this gives that the sub - diamond of regime ( i ) is within the limits @xmath262 , and the sub - diamond of regime ( iii ) is within the limits @xmath263 ( assuming @xmath264 ) .
however , since the virtual levels @xmath265 and @xmath266 are not truly stable , the regimes ( ii ) and ( iii ) show interference patterns that are different than the analyzed in the previous section , as we describe below . regime ( i ) : _ antisymmetric resonances_. if the nearest ( in energy scale ) virtual bath mode at @xmath265 is never reached within the driving interval @xmath267 , the behavior is the same as the one analyzed previously for the ohmic bath in figs.[las](b ) and [ las2](b ) .
the only difference is that the ar interference pattern is now within a smaller sub - diamond , since it is limited by the condition @xmath268 within the driving interval . regime ( ii ) : _ symmetric resonances_. in this case , one of the virtual crossings of @xmath269 with @xmath265 is reached within the driving interval @xmath267 .
therefore , transitions from the @xmath269 level to the virtual level at @xmath265 are possible .
since this later virtual level is unstable , the system decays to the ground state at @xmath270 . by going repeatedly through this process during the periodic driving ,
population is pumped from the @xmath269 level to the ground state at @xmath270 , providing a ` cooling ' mechanism .
the transitions via the unstable mode at @xmath265 lead to a fast relaxation of the system , before the slow mechanisms of bath mediated population inversion , discussed in sec.iiic , can take place .
this impedes the dynamic transition to antisymmetric resonances . in this way
, the interference pattern with symmetric resonant lobes remains _ in the asymptotic regime_. this is shown in fig.[lasp2](a ) , which corresponds to an enlarged section of the regime ( ii ) of fig.[lasp](a ) .
for the ohmic bath .
( a ) for @xmath248 .
( b ) for @xmath249 , title="fig : " ] for the ohmic bath .
( a ) for @xmath248 .
( b ) for @xmath249 , title="fig : " ] regime ( iii ) : _ sideband resonances_. in this case , the two virtual crossings of @xmath223 with @xmath265 are reached within the driving interval @xmath267 .
this makes possible to access also the @xmath266 level through landau - zener transitions at @xmath60 , which allows for new `` sideband '' resonances involving the @xmath265 and @xmath266 levels , in addition to the direct resonances at @xmath271 .
when @xmath272 , called a blue sideband resonance,@xcite the qubit resonates between the @xmath270 and the @xmath266 levels .
however , the @xmath266 virtual level is unstable and it decays to the @xmath269 level . in this way
, the ac drive is continuously pumping population from the ground state to the excited state at @xmath269 , leading to full inversion of the qubit population : @xmath273 .
when @xmath274 , called a red sideband resonance,@xcite the qubit resonates between the @xmath269 and the @xmath265 levels . in this case , the @xmath265 virtual level decays fast to the @xmath270 level , and thus the ac drive is continuously pumping population from the @xmath269 level to the ground state , leading to @xmath275 . in fig.[lasp2](b ) we can see in detail the sideband resonance patterns that characterize this regime , with alternating @xmath276 and @xmath277 .
this type of resonances have been analyzed in ref .
, within a perturbative approach for @xmath278 . in that case , only the regime ( iii ) is realized . on the other hand , when @xmath279 the three regimes described above are possible . to observe the subdivision of the first diamond in the three regimes , the crossing of @xmath269 with @xmath265
should occur before the avoided crossing of @xmath269 with @xmath280 .
this means that the condition @xmath261 is required , or @xmath281 . for the typical flux qubit parameters considered in this work
, this later condition corresponds to @xmath282 .
the squid detectors used in the measurements of refs .
have typically @xmath283 , which is @xmath284 in our normalized units .
therefore in the case of these lzs interferometry experiments , the effects of the resonant mode at @xmath0 are negligible , and everything is within the regime ( i ) of asymptotic asymmetric resonances . on the opposite side , the flux qubits studied in ref .
have @xmath285 , which situates them deeply in the case of the regime ( iii ) .
in fact , the @xmath286 blue and red sideband resonances have been observed in ref .. however , in these devices the oscillator mode of the squid detector is strongly coupled to the qubit , while eq .
( [ jsb ] ) is valid for weak coupling .
a full analysis in this case has to consider a quantum oscillator with frequency @xmath0 coupled to the qubit within the system hamiltonian , see refs .. in any case , our results show that it will be interesting to perform experiments on lzs interferometry using squid detectors with low @xmath0 and weakly coupled to the qubit , to observe the three regimes shown in figs.[lasp ] and [ lasp2 ] .
to summarize , by performing a realistic modeling of the flux qubit we were able to analyze the time dependence of the lzs interference patterns ( as a function of ac amplitude and dc detuning ) taking into account decoherence and relaxation .
we found an important difference between the lzs patterns observed for the time scale of current experiments and those for the asymptotic long time limit : a symmetry change in their structure .
this is a dynamic transition as a function of time from a lzs pattern with nearly symmetric multiphoton resonance lobes to antisymmetric multiphoton resonances .
this transition is observable only when driving the system for very long times , after full relaxation with the bath degrees of freedom ( @xmath175 ) . the large time scale separation , @xmath287 , present in the device of ref .
explains why in their case the asymptotic lzs pattern is beyond the experimental time window
. it will be interesting if experiments could be carried out for longer driving times in this device .
for example , measurements of curves of @xmath149 at growing time scales near a multiphoton resonance , could show the transition from symmetric to antisymmetric behavior .
another interesting finding is the dependence of the lzs interference patterns on the frequency @xmath0 of the squid detector .
different types of lzs interference patterns can arise , depending on the magnitude of @xmath0 .
in particular , we showed that the resonant mode at @xmath0 can impede the dynamic transition when @xmath0 is of the order of the qubit gap , in the regime ( ii ) discussed in sec.iiid . in principle , the frequency @xmath0 can be varied ( in a small range ) by varying the driving current of the squid detector @xcite or with a variable shunt capacitor .
we acknowledge discussions with w. oliver and s. kohler .
we also acknowledge financial support from cnea , uncuyo ( p 06/c455 ) conicet ( pip11220080101821,pip11220090100051,pip 11220110100981,11220150100218 ) and anpcyt ( pict2011 - 1537 , pict2014 - 1382 ) .
since the fq of eq.([ham - sys ] ) is driven with a magnetic flux @xmath288 , the hamiltonian is time periodic @xmath289 , with @xmath290 . in this case
, it is convenient to use the floquet formalism , that allows to treat periodic forces of arbitrary strength and frequency .
@xcite according to the floquet theorem,@xcite the solutions of the schrdinger equation are of the form @xmath291 where @xmath292 are time independent coefficients and @xmath130 are the so - called quasienergies .
the floquet states @xmath129 are time - periodic , @xmath293 and satisfy the eigenvalue equation @xmath294 where @xmath295 is defined as the floquet hamiltonian .
the time periodicity of @xmath296 and @xmath129 makes convenient to use the fourier representation , @xmath297
|\alpha_{q}\rangle= \varepsilon_\alpha |\alpha_{k } \rangle \label{eqfloq1}\ ] ] with @xmath298 the identity . for the driven fq ,
we write @xmath299 with @xmath300 the time independent part of the hamiltonian of eq.([ham - sys ] ) , and @xmath301 \sin(2\pi f_{dc } + 2\varphi_l)+\nonumber\\ & & \!\!\!2\alpha e_j\sin^2[\pi f_{ac}\sin{(\omega_{0 } t)}]\cos(2\pi f_{dc } + 2\varphi_l)\;.\nonumber\\ \label{vt1}\end{aligned}\ ] ] in the energy eigenbasis of @xmath302 , given by @xmath303 , the eq.([eqfloq1 ] ) can be written as @xmath304\langle m|\alpha_q\rangle= \varepsilon_\alpha \langle n|\alpha_k\rangle \label{floq - eig}\ ] ] the static eigenvalues @xmath38 and eigenstates @xmath306 are obtained by numerical diagonalization of @xmath302 , using @xmath307-periodic boundary conditions on @xmath308 and a discretization grid of @xmath309 ( with @xmath310).@xcite then , the @xmath311 are evaluated from eqs .
( [ flo9 ] ) and ( [ vt1 ] ) , where the matrix elements @xmath312 and @xmath313 have been calculated using the obtained eigenstates @xmath306 . in order to solve the floquet eigenvalue problem numerically
we have to truncate the eq .
( [ floq - eig ] ) both in the fourier indices @xmath314 and the in the number of energy levels of @xmath315 considered.@xcite the truncated eigenproblem is of dimension @xmath316 where @xmath317 is defined by the maximum value of the fourier index and @xmath318 by the number of levels considered in the diagonalization of @xmath315 .
the obtained floquet states @xmath319 and quasienergies @xmath130 contain all the information to study the quantum dynamics of the system described above .
an alternative method , which is more efficient for large @xmath318 , is to consider the time evolution operator @xmath320 , which in the floquet representation can be expanded as @xmath321 since @xmath322 , the floquet state @xmath129 is an eigenvector of @xmath323 with eigenvalue @xmath324 .
therefore , it is possible to calculate the floquet states and quasienergies from the diagonalization of @xmath323 .
numerically , one needs to compute the evolution operator @xmath325 within a period , for @xmath326 . taking discretized time steps of length @xmath327
, we use the second - order trotter - suzuki approximation @xmath328 for times @xmath329 , and we compute the product @xmath330 for @xmath331 , starting with @xmath332 .
the floquet states are then obtained as eigenvectors of @xmath333 .
we diagonalize numerically the hermitian matrix @xmath334 , solving @xmath335 , where @xmath336 and @xmath337 , the floquet states at any time are then calculated as @xmath338 , and their fourier components @xmath339 can be obtained using a fast fourier transform routine .
we find that for @xmath340 this numerical procedure is more efficient than the direct diagonalization of eq.([floq - eig ] ) .
experimentally , the probability of having a state of positive or negative persistent current in the flux qubit is measured.@xcite the probability of a positive current measurement ( `` right '' side of the double - well potential ) can be calculated , for @xmath24 , integrating the probability @xmath341 within the subspace with @xmath342:@xcite @xmath343 where @xmath344 and @xmath345 . for later generalizations ,
it is better to define the projector corresponding to a positive current measurement : @xmath346 in terms of this operator , we have @xmath347 . for an initial condition @xmath348 at @xmath349
, we can express @xmath98 in the floquet basis as @xmath350 with @xmath351 . in experiments the initial time , or equivalently the initial phase of the driving field seen by the system in repeated realizations of the measurement , is not well defined .
then , the quantities of interest are the probabilities averaged over initial times , @xmath349.@xcite using the properties of the floquet functions , the average over the initial phase time @xmath349 gives @xmath352 and we obtain , @xmath353}\ ] ] it is worth noticing that averaging over the initial phase time of the driving field is equivalent to defining a density matrix @xmath354 which , due to the average over @xmath349 , is diagonal in the floquet basis , @xmath355 .
finally , we average within a period over the observation time @xmath193 , to obtain a time independent `` stationary '' probability @xmath356 . after defining the time - averaged projector in the floquet basis as @xmath357
we obtain the simple result @xmath358 with @xmath359 .
numerically , after diagonalization of @xmath315 we compute the matrix elements of the projector @xmath360 , with @xmath361.@xcite then , for each @xmath78 and @xmath76 the floquet states and quasienergies are obtained , and the coefficients @xmath362 are evaluated ( with @xmath363 ) . of a state with positive loop current as a function of @xmath75 for the undriven qubit ( red squares ) and for the driven qubit ( black circles ) with @xmath364 and @xmath365 .
the flux detuning @xmath75 is normalized by @xmath187 , such that the @xmath83-photon resonances are at @xmath188 .
( b ) floquet quasienergies ( in units of @xmath14 ) as a function of @xmath75 for the same case as in ( a ) .
( c ) floquet quasienergies ( in units of @xmath14 ) as a function of @xmath76 for the @xmath286 resonant state at @xmath366 [ corresponding to black square in panel ( a ) ] .
( d ) floquet quasienergies ( in units of @xmath14 ) as a function of @xmath76 for an out of resonance state at @xmath367 [ corresponding to blue diamond in panel ( a ) ] .
device parameters of the flux qubit are @xmath42 and @xmath33 . ] as an example , we calculate @xmath368 for the driven fq in the two - level regime , as described by the hamiltonian @xmath369 of eq.([htls ] ) . at zero temperature and in the absence of driving ( @xmath227 )
the isolated fq is in the ground state . in this case , the probability @xmath149 is simply the projection of the ground state on the subspace of positive persistent current , and we have @xmath370 .
for @xmath371 we have @xmath372 except near @xmath373 where @xmath374 , as can be seen in fig.[qef0](a ) .
( while for @xmath375 , we have @xmath376 , since the ground state has the loop current in the opposite direction in this case ) . in the presence of an ac drive , @xmath377
, we calculate the time averaged @xmath378 , following the procedure discussed above .
the time averaged @xmath378 vs. @xmath379 shows dips corresponding to @xmath111photon resonances . as shown in fig.[qef0](a ) .
these @xmath83-resonances are at @xmath380 , which is equivalent to @xmath84 , after defining @xmath82 .
the @xmath83-resonances are at @xmath84 . in the floquet picture , these resonances correspond to avoided crossings of the floquet quasienergies@xcite as a function of @xmath75 as we illustrate in fig.[qef0](b ) .
when increasing @xmath76 , we see that for a @xmath83-resonance the quasienergies have a small gap , @xmath381 ( with the difference @xmath382 defined modulo @xmath78 ) . on the other hand , for a value of @xmath75 away of a resonance
the quasienergies maintain a finite gap ( compared to @xmath78 ) as a function of @xmath76 , see fig.[qef0](c ) and ( d ) .
experimentally , the system is affected by the electromagnetic environment that introduces decoherence and relaxation processes .
a standard theoretical model to study enviromental effects is to couple the system bilinearly to a bath of non - interacting harmonic oscillators with masses @xmath383 , frequencies @xmath384 , momenta @xmath385 , and coordinates @xmath386 , with the coupling strength @xmath387.@xcite the total hamiltonian of system and bath is then given by @xmath388 where @xmath389 is the time - periodic hamiltonian of the system , @xmath390 is the hamiltonian that describes a bath of harmonic oscillators and @xmath391 its system - bath coupling hamiltonian term , with @xmath394 the operator of the system that couples to the bath .
the bath degrees of freedom are characterized by the spectral density @xmath395 it is further assumed that at time @xmath92 the bath is in thermal equilibrium and uncorrelated to the system .
then , the full density matrix @xmath396 has at initial time the form @xmath397 , where @xmath398 is the density matrix of the system and @xmath399 is the bath temperature . after expanding the density matrix of the system in the time - periodic floquet states
@xmath400 the born ( weak coupling ) and markov ( fast relaxation ) approximations for the time evolution of @xmath138 are performed . in this way , the floquet - markov master equation is obtained@xcite @xmath401 the first term in eq.([fbr1 ] ) represents the nondissipative dynamics and the influence of the bath is described by the time - dependent rate coefficients @xmath402 with @xmath403 the nature of the bath is encoded in the coefficients @xmath404 with @xmath405 and @xmath406 , and defining @xmath407 for @xmath408 .
the system - bath interaction is encoded in the transition matrix elements @xmath409 in the floquet basis in the case of the driven fq , the system hamiltonian is @xmath411 and the bath degrees of freedom couple with the system variable @xmath412 since the dominating source of decoherence is flux noise ( see ref . ) .
thus , after taking @xmath413 and in terms of the eigenbasis of @xmath414 , we have to compute @xmath415 considering that the time scale @xmath196 for full relaxation satisfies @xmath416 , the transition rates @xmath417 can be approximated by their average over one period @xmath177 , @xcite @xmath418 , obtaining @xmath419 the rates @xmath420 can be interpreted as sums of @xmath421-photon exchange terms .
this formalism has been extensively employed to study relaxation and decoherence for time dependent periodic evolutions in double - well potentials and in two level systems.@xcite here we use it to model the ac driven fq , considering the full multilevel hamiltonian of eq .
( [ ham - sys]).@xcite to calculate the time dependence of @xmath138 it is convenient to work in the superoperator formalism of the so - called liouville space.@xcite we write , @xmath425 with @xmath426
. then we change notation rewriting the @xmath427 matrix @xmath428 as an @xmath429 vector represented as the ket @xmath430 , and the @xmath431 `` supermatrix '' @xmath179 as the operator @xmath432 acting on this linear space , where the inner product is defined as @xmath433 .
in particular , for the identity matrix @xmath298 we have @xmath434 , the later equality corresponding to the normalization of @xmath428 , which is a conserved quantity . on the other hand ,
the norm of the vector @xmath430 is @xmath435 . in this notation
, we can rewrite the floquet - markov equation as @xmath436 the superoperator @xmath437 is non - hermitian and has left and right eigenvectors with complex eigenvalues @xmath438 , @xmath439 which are mutually orthogonal , @xmath440
. in general , the number of independent eigenvectors of @xmath432 can be less than the dimensionality of @xmath432 . a formal solution of eq.([eqfm ] ) for @xmath441
can be obtained using a similarity transformation to the jordan normal form of @xmath432.@xcite in the cases considered in this work we found numerically that it was always possible to diagonalize @xmath432 , in which case the solution of eq.([eqfm ] ) can be expressed as @xmath442 from where we can calculate numerically @xmath443 .
the probability of a positive current measurement is then obtained combining eq.([p+t ] ) with eq.([rhot1 ] ) .
the asymptotic state @xmath444 satisfies , @xmath445 therefore , the asymptotic state can be constructed from the right - eigenvectors @xmath446 of @xmath432 with eigenvalue @xmath447 ( _ i.e. _ , the kernel of @xmath432 ) .
if @xmath448 is non - degenerate , then the asymptotic state is unique and independent of the initial condition @xmath449 . in the cases considered in this work we found , within the numerical accuracy , that the eigenvalue @xmath447 was non - degenerate , and so the asymptotic state was given by the eigenvector @xmath450 , and then @xmath451 .
the time independence of the asymptotic @xmath452 , implies that quantities like @xmath149 in eq([p+t ] ) , become time - periodic with period @xmath177 in the asymptotic state .
therefore , it is convenient to calculate the asymptotic @xmath98 averaged over one period as @xmath453 it is also clear from eq.([rhot1 ] ) that information about the relaxation and decoherence rates is contained in the non - zero eigenvalues @xmath438 of @xmath432.@xcite on one hand , the relaxation rates are given by the negative real eigenvalues of @xmath432 , where the long time relaxation rate @xmath454 is given by the minimum of @xmath455 ( excluding @xmath447 ) .
on the other hand , the decoherence rates are given by the negative real parts of the complex conjugate pairs of eigenvalues of @xmath432 .
it is well known that the density matrix becomes diagonal in the energy basis for times larger than the relaxation time in undriven systems.@xcite in this case , the decay time of the offdiagonal @xmath456 defines the decoherence time @xmath457 between the eigenstates @xmath38 and @xmath458 . for time scales
@xmath459 , relaxation is described by the pauli rate equation for the populations of the energy levels @xmath460 . in the case of a system with a time periodic drive
, it is usually assumed that for large times the density matrix becomes approximately diagonal in the floquet basis.@xcite more precisely , this approximation can be justified when @xmath461 , which is fulfilled for very weak coupling with the environment and away from resonances , see ref .. from fig.[qef0](c ) and ( d ) it is clear that this condition will be easily satisfied in the offresonant case considered here , where the floquet gap @xmath462 is large .
on the other hand at an near a resonance where @xmath463 this condition can not be fulfilled , unless the system - bath coupling is extremely small.@xcite as a function of time .
black lines : off - diagonal matrix element @xmath464 in the basis of eigenstates of the undriven hamiltonian .
red lines : off - diagonal element @xmath465 in the floquet basis .
( a ) for a resonant state at @xmath466 driven with @xmath364 and @xmath365 .
( b ) for an off - resonant state at @xmath367 driven with @xmath107 and @xmath365 .
insets in ( a ) and ( b ) show @xmath465 at large times . ] in fig.[foff ] we show the time evolution of matrix elements of @xmath428 calculated in the eigenbasis of @xmath414 and in the floquet basis , both for an offresonant and for a resonant case . to clarify notation , for the @xmath414 eigenbasis we use latin indices @xmath467 with @xmath468 ordered for increasing eigenenergy @xmath469 .
for the floquet basis we use greek indices @xmath470 with @xmath471 , and ordered such that @xmath472 corresponds to the state that in the limit @xmath230 maps to the ground state ( with index @xmath473 ) ; @xmath474 corresponds to the state that in the limit @xmath230 maps to the first excited state ( with index @xmath475 ) , and so on .
the time dependence of the offdiagonal elements @xmath476 in the the @xmath414 eigenbasis , are shown in fig.[foff](a ) and fig.[foff](b ) for the resonant and offresonant cases , respectively .
( here we consider the flux qubit in contact with an ohmic bath with @xmath477 ) .
we see that for long times @xmath478 goes to a finite non - zero value that can not be neglected , showing explicitly that the driven system density matrix is not diagonal in the eigenenergy basis
. therefore approaches based on the use of the pauli rate equation in the eigenenergy basis will not be correct for the analysis of the asymptotic long time behavior . in the case of the floquet basis
, we see in fig.[foff](a ) and ( b ) that the offdiagonal @xmath465 after having oscillations at short time scales , decreases exponentially to very low values for long times .
we find that @xmath479 for the offresonant case ( while diagonal elements @xmath480 ) .
this confirms that it is a good approximation to neglect the @xmath465 at long times in this case . on the other hand , in the resonant case we find @xmath481 .
thus , in this case neglecting @xmath465 is not a good approximation as in the offresonant case .
the results reported in the main body of the paper correspond to the solution of the full floquet - markov equation , eq . [ eqfm ]
. however , we have verified that most of our results ( including the dynamic transition discussed in sec.iii ) are accurately reproduced by the approximation that assumes an asymptotic density matrix diagonal in the floquet basis .
assuming that the density matrix becomes diagonal in the floquet basis , one can separate the dynamics of the diagonal and the off - diagonal density matrix .
the off - diagonal part is dominated by the dependence @xmath482 \,\rho_{\alpha\beta}\;\;\;\ ; \alpha\not=\beta\ ] ] in this approximation , the decoherence rate between the @xmath483 and the @xmath484 floquet state , is given by @xmath485 .
the dynamics for the diagonal part of the density matrix gives a rate equation for the population of the floquet states @xmath486 : @xmath487 where @xmath488 , after eq .
( [ rates ] ) .
it is simple to solve the above rate equation when we restrict to two levels . with two floquet states @xmath489 ,
the asymptotic populations are @xmath490 , @xmath491 ; and the relaxation rate is @xmath492 . using eq.([rates ] ) , we can decompose the relaxation rate as a sum of terms that describe virtual n - photon transitions:@xcite @xmath213 with @xmath493 d. m. berns , m. s. rudner , s. o. valenzuela , k. k. berggren , w. d. oliver , l. s. levitov , and t. p. orlando , nature * 455 * , 51 ( 2008 ) ; w d. oliver and s. o. valenzuela , quantum inf . process * 8 * , 261 ( 2009 ) .
a. izmalkov , s. h. w. van der ploeg , s. n. shevchenko , m. grajcar , e. ilichev , u. hbner , a. n. omelyanchouk , and h .-
meyer1 , phys .
* 101 * , 017003 ( 2008 ) .
m. sillanpaa , t. lehtinen , a. paila , y. makhlin , and p. hakonen , phys .
lett . * 96 * , 187002 ( 2006 ) . c. m. wilson , t. duty , f. persson , m. sandberg , g. johansson , and p. delsing , phys .
lett . * 98 * , 257003 ( 2007 ) .
g. z. sun et al , appl .
. lett . * 94 * , 102502 ( 2009 ) .
guozhu sun , xueda wen , bo mao , yang yu , jian chen , weiwei xu , lin kang , peiheng wu , and siyuan han , phys .
b 83 , 180507(r ) ( 2011 ) .
yiwen wang , shanhua cong , xueda wen , cheng pan , guozhu sun , jian chen , lin kang , weiwei xu , yang yu , and peiheng wu phys .
b 81 , 144505 ( 2010 ) .
s. e. de graaf , j. leppkangas , a. adamyan , a. v. danilov , t. lindstrm , m. fogelstrm , t. bauch , g. johansson , and s. e. kubatkin phys .
111 , 137002 ( 2013 ) .
j. stehlik , y. dovzhenko , j. r. petta , j. r. johansson , f. nori , h. lu , and a. c. gossard phys .
b 86 , 121303(r ) ( 2012 ) .
gang cao , hai - ou li , tao tu , li wang , cheng zhou , ming xiao , guang - can guo , hong - wen jiang , and guo - ping guo nat comms 4 1401 ( 2013 ) .
e. dupont - ferrier , b. roche , b. voisin , x. jehl , r. wacquez , m. vinet , m. sanquer , and s. de franceschi phys .
110 , 136802 ( 2013 ) .
runan shang , hai - ou li , gang cao , ming xiao , tao tu , hongwen jiang , guang - can guo , and guo - ping guo appl .
103 * 162109 ( 2013 ) .
p. nalbach , j. knrzer , and s. ludwig phys .
b 87 , 165425 ( 2013 ) f. forster , g. petersen , s. manus , p. hnggi , d. schuh , w. wegscheider , s. kohler , and s. ludwig phys .
112 * , 116803 ( 2014 ) j.i .
colless , x.g .
croot , t.m .
stace , a.c .
doherty , s.d .
barrett , h. lu , a.c .
gossard , and d.j .
reilly nat comms 5 ( 2014 ) .
g. granger , g. c. aers , s. a. studenikin , a. kam , p. zawadzki , z. r. wasilewski , and a. s. sachrajda phys .
b 91 , 115309 ( 2015 ) .
georg heinrich , j. g. e. harris , and florian marquardt phys .
a 81 , 011801(r)(2010 ) ; huaizhi wu , georg heinrich and florian marquardt , new journal of physics 15 , 123022 ( 2013 ) .
s. gasparinetti , p. solinas , and j. p. pekola phys .
107 , 207002 ( 2011 ) , xinsheng tan , dan - wei zhang , zhentao zhang , yang yu , siyuan han , and shi - liang zhu phys .
112 , 027001 ( 2014 ) , junhua zhang , jingning zhang , xiang zhang , and kihwan kim , phys .
rev . a 89 , 013608 ( 2014 ) . j. e. mooij , t.p.orlando , l.s .
levitov , l. tian , c.h .
van der wal and s. lloyd , science * 285 * , 1036 ( 1999 ) ; t.p.orlando , j.e .
mooij , l. tian , c.h .
van der wal , l.s .
levitov , s. lloyd , j.j .
mazo , phys .
b * 60 * , 15398 ( 1999 ) .
we diagonalized the hamiltonian using the representation in terms of @xmath308 and periodic boundary conditions in the wave function @xmath494 .
the discretization grid is @xmath309 , with @xmath310 , and the lowest eigenvalues and eigenvectors are obtained with a lanczos algorithm , employing the arpack library
. for low energies , the potential @xmath26 is approximately separable in the longitudinal [ @xmath495 and transverse [ @xmath496 directions near the two minima .
therefore , at low energies the first eigenstates can be assigned longitudinal and transverse quantum numbers .
the black curves in fig.[fig - espec ] correspond to eigenstates with the lowest transverse mode , _
i.e. _ wave functions symmetric in @xmath497 .
the blue curves in fig.[fig - espec ] correspond to eigenstates with higher transverse modes , not necessarily symmetric in @xmath497 .
the driving in @xmath48 is along the longitudinal direction and therefore conserves the parity of the transverse modes .
the avoided crossings indicated as @xmath498 in fig.[fig - espec ] have a negligibly small gap @xmath499 .
sriram ganeshan , edwin barnes , and s. das sarma phys . rev .
. 111 130405 ( 2013 )
. s. kohler , t. dittrich and p. hnggi , phys .
e. * 55 * , 300 ( 1997 ) .
s. kohler , r. utermann , p. hnggi , and t. dittrich , phys .
e. * 58 * , 7219 ( 1998 ) .
s. gasparinetti , p. solinas , s. pugnetti , r. fazio , and j. pekola phys .
. 110 150403 ( 2013 ) .
vera gramich , simone gasparinetti , paolo solinas , and joachim ankerhold phys .
113 , 027001 ( 2014 ) .
w. thomas pollard and richard a. friesner , j. chem .
100 , 5054 ( 1994 ) ; w. thomas pollard , anthony k. felts , and richard a. friesner , advances in chemical physics , volume xciii , edited by i. prigogine and stuart a. rice . 1996
john wiley & sons , inc . | we study landau - zener - stuckelberg ( lzs ) interferometry in multilevel systems coupled to an ohmic quantum bath .
we consider the case of superconducting flux qubits driven by a dc+ac magnetic fields , but our results can apply to other similar systems .
we find a dynamic transition manifested by a symmetry change in the structure of the lzs interference pattern , plotted as a function of ac amplitude and dc detuning .
the dynamic transition is from a lzs pattern with nearly symmetric multiphoton resonances to antisymmetric multiphoton resonances at long times ( above the relaxation time ) .
we also show that the presence of a resonant mode in the quantum bath can impede the dynamic transition when the resonant frequency is of the order of the qubit gap .
our results are obtained by a numerical calculation of the finite time and the asymptotic stationary population of the qubit states , using the floquet - markov approach to solve a realistic model of the flux qubit considering up to 10 energy levels . |
two of the most widely used tracers of recent star formation are the nebular h@xmath0 recombination line and the ultra - violet ( uv ) continuum .
the h@xmath0 arises from recombination of gas ionized by photons from massive stars ( @xmath7 15 m@xmath6 ) and is expected to be observed over the typical lifetimes of extremely massive stars ( @xmath4 5 myr ) .
the uv continuum is due to non - ionizing photospheric emission from stars with @xmath8 @xmath7 3 m@xmath6 , which have lifetimes @xmath4 300 myr . in tandem ,
these two integrated tracers provide leverage on current and recent star formation in both nearby and distant galaxies ( see @xcite and references therein ) . in principle , observed h@xmath0 and uv luminosities should yield consistent measures of star formation . under the assumption of a constant sfh over a sufficiently long timeline ( e.g. , @xmath1 1 gyr ) ,
the expected ratio of h@xmath0 and uv star formation rates ( sfrs ) should be constant with respect to both time and environment ( see @xcite and references therein ) , assuming a fully populated and universal imf , solar metallicity , all lyman continuum photons ionize hydrogen , and with no attenuation due to dust .
deviations from this fiducial ratio would suggest that one or more of the underlying assumptions are not correct .
a number of studies have demonstrated discrepancies in sfrs as measured by h@xmath0 and uv luminosities ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
recently , several studies have unambiguously shown that the observed discrepancy in h@xmath0 and uv sfr indicators is systematic , such that the observed ratio of h@xmath0-to - uv flux declines with decreasing galaxy luminosity ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
lcccccc ugca292 & 5.65 & 6.91 & -11.42 & -2.755 & 0.031 & 1 + ugc5364 & 5.82 & 7.24 & -11.40 & -4.135 & -0.739 & 1 + ugc8091 & 5.84 & 7.54 & -12.07 & -2.700 & -0.007 & 1 + ugca438 & 6.28 & 7.65 & -12.35 & -4.316 & -1.668 & 2 + ugc8833 & 6.44 & 7.60 & -12.42 & -3.331 & -0.468 & 1 + ugc4483 & 6.50 & 7.72 & -12.86 & -2.446 & 0.016 & 2 + ugc9128 & 6.51 & 7.56 & -12.39 & -3.970 & -0.948 & 1 + cgcg269 - 049 & 6.52 & 7.88 & -12.35 & -3.120 & -0.290 & 1 + kkh37 & 6.53 & 7.56 & -11.67 & -3.818 & -0.319 & 2 + ugca281 & 6.54 & 8.32 & -13.50 & -1.391 & 0.369 & 1 [ tab1 ] despite extensive research , there remains no consensus for the cause of this trend .
independent studies have verified that factors such as assumed metallicity and choice of stellar models can not be responsible for the observed trend ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , while investigations of other contributing factors , including non - constant sfhs , a variable imf , and ionizing photon leakage have yet to yield conclusive results . in this paper
, we undertake a focused investigation on the impact of non - constant sfhs on observed h@xmath0-to - uv ratios .
our main objective is to find a set of simple sfh models ( i.e. , periodic bursts of star formation superimposed on a baseline constant sfr ) that are well matched to the observed distribution of h@xmath0-to - uv sfr ratios , while being consistent with other available data . to do this , we compare the observed distribution of h@xmath0-to - fuv ratios to the predicted distributions from a set of simple sfh models , in bins of galaxy stellar mass . using the results of this comparison
, we examine the ability of the highest probability model sfhs to explain the observed trend in h@xmath0-to - fuv flux ratios versus stellar mass , @xmath2-band surface brightness , h@xmath0 luminosity , and @xmath9 .
in addition , we compare sfh model parameters and predictions with previous studies that have analyzed trends in h@xmath0-to - uv ratios using different sfhs or imf assumptions .
we conclude the paper by discussing the strengths and limitations of simple model sfhs and suggest specific ways to improve future sfh modeling efforts .
lccc|cc|ccc 0.50 & -0.14 & 0.29 & 20 & -0.32 & 0.16 & -0.06 & 0.05 & 7 + -0.25 & -0.09 & 0.20 & 26 & -0.28 & 0.13 & -0.03 & 0.08 & 22 + -0.75 & -0.13 & 0.22 & 44 & -0.22 & 0.15 & -0.06 & 0.10 & 21 + -1.25 & -0.13 & 0.22 & 51 & -0.19 & 0.15 & -0.06 & 0.14 & 36 + -1.75 & -0.18 & 0.17 & 58 & -0.24 & 0.26 & -0.14 & 0.24 & 38 + -2.25 & -0.27 & 0.22 & 56 & -0.38 & 0.45 & -0.26 & 0.44 & 40 + -2.75 & -0.51 & 0.25 & 31 & -0.38 & 0.45 & -0.26 & 0.44 & 15 + -3.50 & -0.55 & 0.57 & 21 & -0.69 & 0.31 & -0.53 & 0.26 & 6 [ samplediffs ] we test our sfh models on a sample of 185 nearby star - forming galaxies , whose observable properties are listed in table [ tab1 ] .
to arrive at this particular sample , we began by considering the 390 galaxies with h@xmath0 and galex fuv observations from the 11 mpc h@xmath0 and uv galaxy survey ( 11hugs ; * ? ? ?
* ; * ? ? ?
we further restricted the sample to those galaxies that also had high quality ancillary data to to allow us correct for the effects of dust and study the h@xmath0-to - fuv ratio trends versus galaxy stellar mass and surface brightness .
specifically , we selected 11hugs galaxies that are also members of the smaller _ spitzer _ local volume legacy survey ( lvl ; e.g. , * ? ? ? * ) .
the lvl sample contains 258 galaxies with comprehensive _ spitzer _ ir imaging .
however , only 59 galaxies in the sample also have homogeneous optical broadband imaging @xcite .
to increase the sample size , we also considered lvl galaxies that fall within the footprint of sdss dr7 @xcite .
there are 126 additional galaxies that met this requirement .
thus , we have a final sample of 185 galaxies with observations of h@xmath0 and galex fuv from 11hugs , _ spitzer _ ir from lvl , and optical broadband imaging from either sdss or lvl . for consistency with the lvl optical imaging
, fluxes measured from sdss images were converted to the johnson filter system using the transformations provided in @xcite . in figure
[ janicesample ] , we show a comparison between the h@xmath0-to - fuv ratios from @xcite ( navy and grey points ) with those of the 185 galaxies considered in this study ( navy points only ) .
the h@xmath0 , fuv , and b - band fluxes , as well as the adopted dust corrections in this figure have all been taken from @xcite .
although similar in most aspects , there are two notable difference between the samples .
first , our subset of 185 galaxies are noticeably deficient in luminous systems ( @xmath9 @xmath1 @xmath1016 ) with h@xmath0-to - fuv ratios significantly greater than the fiducial .
second , our sample contains fewer low luminosity systems than the larger sample of @xcite . in both cases ,
the limiting factor is the lack of sufficient ancillary data . in the case of the low luminosity galaxies , nearly all that have been excluded from the present study lack sufficiently homogenous optical broadband coverage . for more luminous galaxies , some lack either ancillary optical imaging or fall outside the boundaries of the lvl survey and therefore lack _ spitzer _ imaging .
we next examine the differences in h@xmath0-to - fuv ratios between the two samples . in table
[ samplediffs ] , we quantify the h@xmath0-to - fuv ratios as a function of h@xmath0 sfr , after applying identical extinction corrections from @xcite ( see [ duscor ] ) .
as expected , the largest difference is seen at high h@xmath0 sfr , where our sample lacks galaxies with high h@xmath0-to - fuv ratios . for @xmath11
$ ] @xmath7 @xmath101.75 , the mean h@xmath0-to - fuv ratios in the present study are systematically @xmath1 0.2 dex lower than those in the sample of @xcite .
similarly , the 1@xmath12 scatter over the same range is @xmath1 0.1 dex lower in the present sample .
however , for galaxies with @xmath11 $ ] @xmath4 @xmath101.75 , the h@xmath0-to - fuv ratios have comparable mean values and scatter , per bin of h@xmath0 sfr . outside of the differences between the samples at high h@xmath0 sfrs ,
the two samples are in general agreement .
this indicates that possible selection effects associated with the smaller sample in this study are minimal .
for accurate comparison of h@xmath0 and fuv sfr indicators , corrections are required for extinction due to both galactic foreground and to internal effects in each galaxy .
we correct for foreground extinction using the maps of @xcite , assuming the milky way extinction law of @xcite . for internal attenuation , we use two types of corrections : the first based on direct ir measurement of reprocessed light and another using correlation - based prescriptions ( e.g. , see appendix d in @xcite ) . for @xmath1 70% of the sample , we have estimates of the total - ir flux based on spitzer / mips far - ir ( fir ) measurements , allowing us to use the ` energy balance ' extinction correction methods derived by @xcite for h@xmath0 and by @xcite for fuv .
these methods derive an intrinsic luminosity from the combination of unobscured ( e.g. , h@xmath0 , fuv ) and obscured ( i.e. , fir continuum ) signatures of star formation .
from the intrinsic luminosity one can correct for the effects of internal dust attenuation on the observed h@xmath0 and fuv fluxes .
for the @xmath1 30% of the sample where fir measurements were not available , we used the same correlation - based attenuation corrections as @xcite , namely a @xmath13-band luminosity prescription for @xmath14 and a dust - law prescription for @xmath15 .
while both methods are anchored to balmer decrement extinction corrections , the self - consistency of the ` energy balance ' method along with the availability of ir for the majority of our sample make it an appealing choice for this study .
more explicitly , the ` energy balance ' extinction correction for @xmath14 from @xcite is : @xmath16 and from @xcite , the correction for the fuv is given as : @xmath17 where the tir luminosity is a mips - based measurement made using the conversion in @xcite : @xmath18 overall , our internal extinction corrections yield similar results to those used by @xcite .
as shown in table [ samplediffs ] , the main difference between the two sets of corrections is seen for the most luminous galaxies in the sample . here , @xcite find the mean h@xmath0-to - fuv ratio to be @xmath1 0.2 - 0.3 dex below the expected fiducial , whereas the energy balance dust corrections place the mean ratios within @xmath1 0.05 dex of the fiducial .
however , for galaxies with low h@xmath0 sfrs , the adopted extinction correction does not drastically influence the mean or scatter in the h@xmath0-to - fuv ratios ; these galaxies have low dust contents , such that extinction corrections have only modest impacts on their luminosities .
we plot the flux ratios for our sample with the energy balance extinction corrections applied in figure [ hasfrmb ] .
although differences in dust corrections are both important and interesting in their own right , analyzing the impact of various extinction corrections on sfr indicators is beyond the scope of the present analysis .
instead , we refer the reader to @xcite , which investigates the reliability and consistency of various h@xmath0 and fuv extinction corrections , including those from @xcite , @xcite , and @xcite
. the stellar mass for each galaxy in the sample was derived by spectral energy distribution ( sed ) fitting as detailed in @xcite , which closely follows the methods detailed in @xcite .
briefly , we fit the observed uv , optical data , and ir luminosities of each galaxy with a suite of @xcite population synthesis models , whose parameters span a range in stellar metallicity , age , exponentially declining sfhs , and dust properties . for each model , we derive a stellar mass from the best fitting normalization of the model sed
. we also construct the cumulative distribution function for the stellar mass from the likelihoods of the normalized model seds for each sample galaxy .
the stellar masses reported here are the medians of the cumulative distribution function . in practice ,
the stellar mass is surprisingly robust , as the effects of dust and age on the optical mass - to - light ratio have correspondingly similar effects on the observed optical color @xcite , while the effects of stellar metallicity on the optical mass - to - light ratios are minimal .
the derived stellar masses for each system are presented in table [ tab1 ] . the h@xmath0-to - fuv ratio as a function of galaxy stellar mass ( figure [ obs_data ] ) shows subtle , but important differences compared to when the ratios are plotted versus h@xmath0 sfr or @xmath9 , as in figure [ hasfrmb ] .
for example , when considering the ratios versus h@xmath0 sfr , we see there are no galaxies with h@xmath0-to - fuv ratios above the fiducial for @xmath19 $ ] @xmath4 @xmath102.5 .
additionally , there are only a few galaxies with h@xmath0-to - fuv ratios near or above the fiducial for systems with @xmath9 @xmath4 @xmath1013 . from these two panels
, one could conclude that galaxies with low luminosities or low current sfrs must have systematically low h@xmath0-to - fuv ratios . however , when plotting the same ratios versus stellar mass , one could reach a different conclusion . as shown in the top panel of figure [ obs_data ] , a number galaxies with masses @xmath4 10@xmath5 m@xmath6 have h@xmath0-to - fuv ratios near or above the fiducial .
recent bursts of sf have little impact on a galaxy s total mass , but can strongly influence the observed h@xmath0 or optical luminosities ( e.g. , * ? ? ? * ) .
thus , considering the data versus total stellar mass minimizes the covariances that affect the h@xmath0-to - fuv ratios when plotted versus luminosity . a similar study conducted by @xcite found a strong trend between @xmath2-band surface brightness ( @xmath20 ) and h@xmath0-to - fuv ratios . to facilitate a direct comparison , we utilize @xmath2-band imaging from either sdss or lvl , and follow the surface brightness measurement procedure described in @xcite .
below , we briefly summarize the methodology .
we first masked foreground stars and background galaxies identically to the method described in @xcite . with these objects removed , we computed the surface brightness profiles using elliptical apertures .
the major and minor axes of a measurement aperture are scaled , preserving ellipticity , such that the ellipse encloses 50% of the total flux .
we then normalized the total flux to the area of the aperture to derive the surface brightness .
particular care was taken to ensure that these scaled measurement apertures , constructed using ellipticity and position data compiled from ned / simbad , were properly shaped and centered .
for consistency with @xcite , we note that the flux percentage choice of 50% rather than 90% does not affect the overall trends in our data , beyond shifting to slightly higher surface brightnesses .
we plot the h@xmath0-to - fuv ratios versus @xmath2-band surface brightness ( @xmath20 ) in the bottom panel of figure [ obs_data ] .
again , we see a decline in h@xmath0-to - fuv ratios toward lower surface brightness systems . in this case , the systematic decline of h@xmath0-to - fuv ratios is not as strong as when the sample is plotted versus h@xmath0 sfr , but is still more evident than when the ratios are considered versus galaxy stellar mass . like all optical luminosities , @xmath2-band fluxes are known to vary due to episodes of recent sf ( e.g. , * ? ? ?
* ) , which likely explains at least some of the correlation between h@xmath0-to - fuv ratios and @xmath20 .
we note that the observed trend between h@xmath0-to - fuv ratio and @xmath2-band surface brightness in our sample is not quite as strong as that presented in @xcite , which we have over - plotted as the red dotted line in figure [ obs_data ] .
we discuss some of the potential reasons for this disagreement in [ previous ] .
lcc @xmath21 @xmath2210.0 & -0.05@xmath23 & 16 + 10.0@xmath24@[email protected] & -0.01@xmath25 & 20 + 9.0@xmath24@[email protected] & -0.06@xmath26 & 55 + 8.0@xmath24@[email protected] & -0.17@xmath27 & 71 + @xmath21 @xmath287.0 & -0.27@xmath29 & 23 [ tab2 ]
the model sfhs we consider in this analysis are simple toy models that are primarily composed of periodic bursts superimposed onto a constant sfr .
given that modest variations in the period , duration , and amplitude of bursts can result in significantly different h@xmath0 , uv , and broadband fluxes , we consider a diverse and extensive set of @xmath1 1500 model sfhs , whose properties and construction we describe below .
we model sfhs that focus on two separate regimes : the ancient epoch ( @xmath22 1 gyr ago ) which has little impact on the h@xmath0-to - fuv ratios and recent sfhs ( @xmath30 1 gyr ago ) which are characterized by a mix of constant sf and bursts .
we have modeled the older stellar populations that dominate a galaxy s total stellar mass using a 1 gyr burst of sf 8 gyr ago to populate the red giant branch and ancient main sequence , as well 1 gyr of constant sfr from 1 - 2 gyr ago to account for the contribution of asymptotic giant branch stars .
the precise epoch of the ancient and intermediate age sfhs do not significantly change any of our analysis . changing the ages of these older bursts results in small variations in the absolute values of @xmath31 , i.e. , @xmath4 0.5 dex . however , because the same ancient and intermediate sfhs are used for all models , there is no effect on the relative values of @xmath31 .
finally , to match findings from measured sfhs in nearby dwarf galaxies ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , we scaled the sfhs such that sf from @xmath22 1 gyr ago accounts for 90% of the total stellar mass formed in each model .
we constructed the recent model by assuming a constant sfr over the past 1 gyr and superimposing bursts of sf .
we assume the bursts to be a 1 myr long episode of constant sf , which we then interpolated onto a finer time grid to track the evolution of h@xmath0 and uv fluxes over longer timescales .
this basis model was then shifted in time , multiplied by the appropriate amplitude , and finally linearly co - added with other basis models to obtain the desired recent sfh .
each model has a specified burst amplitude ( a ; the ratio of maximum sfr to constant sfr ) , burst duration ( d ) , and period between bursts ( p ) .
we additionally characterize our models by the quantity ( @xmath32@xmath33@xmath34)/@xmath35 , which is the ratio of stellar mass formed during a single burst to the mass formed assuming the baseline sfr over one period .
selected permutations of a ( 2 , 5 , 10 , 15 , 20 , 30 , 40 , 50 , 75 , 100 , 150 ) , d ( 3 , 6 , 12 , 18 , 25 , 30 , 40 , 50 , 60 , 75 , 100 , 150 , 200 myr ) , and p ( 5 , 10 , 15 , 20 , 25 , 30 , 40 , 50 , 60 , 75 , 100 , 125 , 150 , 175 , 200 , 225 , 250 myr ) combined with the constraint @xmath36 resulted in 1441 different model sfhs .
we added 25 more models in which we allowed the sfr between bursts to go to zero , i.e. , the burst amplitude is infinite .
we also consider a constant sfh model ( i.e. , a @xmath37 1 ) .
many of the models we consider fall under the literature parlance of ` bursting ' or ` gasping ' sfhs ( e.g. , * ? ? ? * ; * ? ? ?
although the definition of both terms is not universal , here we consider a ` bursting ' model to have @xmath38 and @xmath39 , while for a ` gasping ' model @xmath32 @xmath1 @xmath35 with the deficit in sf equal to 1/a . that is , a gasp is essentially a decrease in the sfr from an otherwise constant rate .
we show a select sample of model sfhs in figure [ sample_sfhs ] .
-to - fuv flux ratio distribution in order of decreasing bins of stellar mass ( ( a ) @xmath21 @xmath2210.0 ,
( b ) 10.0@xmath24@[email protected] , ( c ) 9.0@xmath24@[email protected] , ( d ) 8.0@xmath24@[email protected] , ( e ) @xmath21 @xmath287.0 ; see table [ tab3 ] ) , as determined by a two sided ks test .
the inset shows the evolution of the h@xmath0-to - fuv sfr ratio as a function of time for each model sfh .
the red , green , blue , magenta , and cyan color coding each refer to the best fit sfh models in the bins of highest to lowest stellar mass , respectively .
this color coding scheme will be used throughout the remainder of the paper in reference to a particular sfh model . ] synthesizing the evolution of h@xmath0 , fuv , and @xmath2-band fluxes was done consistently with the construction of the model sfhs .
a 1 myr long burst of sf was input into the synthesis code of @xcite .
the resulting spectra were convolved with the appropriate filter response functions , and the resulting fluxes were linearly co - added identically to the sfhs . in calculating these integrated quantities , we selected the padova stellar evolution models @xcite , solar metallicity , and a fully populated chabrier imf with mass limits of 0.1 and 100 m@xmath6 . for consistency with the observations , we convolved the uv spectrum with the galex fuv filter response function .
the h@xmath0 flux is modeled as the number of ionizing photos divided by @xmath40 , the expectation value for case b recombination .
we do not include losses due to the leakage or absorption of ionizing photons by dust within the strmgren sphere . in these models ,
the h@xmath0 and fuv fluxes provide for consistent measures of sf when averaged over the course of 1 gyr , i.e. , the ratio of the sums , @xmath41(h@xmath0)/@xmath41(uv ) , is constant , even though many individual points may deviate from this average . throughout the paper , we assume a fiducial value of h@xmath0-to - fuv ratios based on the scenario in which a galaxy which has been forming stars at a constant rate over its entire lifetime . for clarify in plotting , we have set the logarithm of the fiducial h@xmath0-to - fuv ratio equal to zero .
consequently , all model and observed h@xmath0-to - fuv ratios have been adjusted appropriately such that @xmath42_{plotted}$ ] @xmath37 @xmath42 + \kappa$ ] , where @xmath43 @xmath1013.17 to derive the model @xmath2-band surface brightness values , we converted the @xmath2-band fluxes to a surface brightness assuming an area of 1 kpc@xmath44 . in practice , this areal normalization will increase for more massive galaxies .
however , as we are only interested in variations in the logarithmic surface brightnesses , altering the areal normalization by a factor of a few has minimal impact on the placement of the models .
overall , this method results in model @xmath2-band fluxes ( and surface brightnesses ) that are consistent with observations of galaxies with these typical sfrs .
uncertainties in the observed h@xmath0 , fuv , and @xmath2-band fluxes are not inherently factored into the modeled fluxes .
as this may contribute to the observed scatter , we applied independent small random shifts based on observational uncertainties in the h@xmath0 and fuv fluxes listed in @xcite and @xcite , to each of the model flux values .
specifically , we drew a set of errors from a randomly generated normal distribution with a mean of 0 and a dispersion of 0.05 mag .
lcccccc @xmath21 @xmath2210.0 & best & 125 & 3 & 10 & 0.24 & 72.3 + & 10th best & 75 & 12 & 2 & 0.32 & 65.6 + & 10th worst & 125 & 12 & 150 & 14.4 & 4.9@xmath3310@xmath45 + 10.0@xmath24@[email protected] & best & 175 & 6 & 5 & 0.17 & 99.9 + & 10th best & 225 & 6 & 5 & 0.13 & 99.6 + & 10th worst & 75 & 3 & 100 & 4.00 & 2.53@xmath3310@xmath46 + 9.0@xmath24@[email protected] & best & 250 & 60 & 5 & 1.20 & 94.3 + & 10th best & 225 & 40 & 5 & 0.89 & 83.4 + & 10th worst & 75 & 3 & 150 & 6.00 & 1.67@xmath3310@xmath47 + 8.0@xmath24@[email protected] & best & 250 & 30 & 30 & 3.60 & 97.5 + & 10th best & 250 & 25 & 40 & 4.00 & 73.4 + & 10th worst & 75 & 12 & 100 & 16.0 & 1.1@xmath3310@xmath48 + @xmath21 @xmath287.0 & best & 200 & 40 & 30 & 6.00 & 99.8 + & 10th best & 150 & 30 & 20 & 4.00 & 95.8 + & 10th worst & 250 & 6 & 2 & 0.05 & 3.6@xmath3310@xmath49 [ tab3 ]
the goal of our analysis is to identify a set of characteristic sfhs that can match the observed distribution of h@xmath0-to - fuv ratios within bins of galaxy stellar masses , without violating other observational constraints . to determine this characteristic sfh
, we compare the observed distribution of h@xmath0-to - fuv ratios with predicted distributions from each of the model sfhs .
we first divide the observational sample into bins of stellar mass , under the assumption that galaxies with comparable stellar masses are likely to share a characteristic sfh .
we chose a simple scheme that consists of five bins with sizes of @xmath50@xmath51 1 .
the resulting division provides a reasonable balance between grouping galaxies of similar masses and populating each bin with an adequate number of data points .
the number of galaxies , median h@xmath0-to - fuv ratio , and scatter in the ratio for each bin are listed in table [ tab2 ] . as a metric for comparison between the models and observed h@xmath0-to - fuv ratios , we have adopted a two - sided kolmogorov - smirnov ( ks ) test . unlike other statistical tests ( e.g. , @xmath52-test , anderson - darling )
, the ks test makes no assumption about the form of the parent distribution of either population , i.e. , they need not be normally distributed .
indeed , the distributions of the h@xmath0-to - fuv ratios are generally complex and difficult to parameterize , making the ks test better suited for this particular analysis . for simplicity
, we will adopt multiples of the typical @xmath12 values to evaluate the quality of the model - data agreement ( e.g. , 1-@xmath12 @xmath1 68% , 2-@xmath12 @xmath1 95% , etc ) .
we also include parameters and ks test probabilities for the 10th best and 10th worst fits to highlight the differences in the model parameters in the high and low probability regimes .
the results of the model - data comparison are summarized in table [ tab3 ] . in each bin of stellar mass ,
we have indicated the highest probability model parameters ( i.e. , @xmath35 , @xmath32 , @xmath34 ) , the value of ( @xmath32@xmath33@xmath34)/@xmath35 , and the ks probability . in general , in each bin of stellar mass , our highest probability models provide reasonable matching to the observed distributions of h@xmath0-to - fuv ratios , as judged by the ks test .
four of the five bins have models that match the data with a probability of @xmath7 94% .
the lowest mass bin ( @xmath21 @xmath287.0 ) and second highest mass bin ( 10.0@xmath24@[email protected] ) have ks probabilities of 99.8% and 99.9% , respectively .
the lowest probability model is in the highest mass bin ( @xmath21 @xmath2210.0 ) , which has a ks probability of 72.3% . despite the simplicity of the sfh models considered
, we find all highest probability models agree with the data at levels of @xmath7 1 - 3 @xmath12 .
examining the highest probability model parameters in each stellar mass bin , we see the sfhs of higher mass systems are generally characterized by constant sfhs with an occasional modest burst . specifically , the two highest mass bins favor models of short duration ( @xmath32 @xmath4 6 myr ) and relatively modest amplitudes ( @xmath34 @xmath4 10 ) .
these sfhs are predominantly constant sfrs , with bursts interspersed @xmath1 5% of the time . the central bin ( 9.0@xmath24@[email protected] )
is best characterized by a long period ( @xmath53 250 myr ) of constant sf and an increase in amplitude by a factor of 5 for a 60 myr duration .
the two lowest mass bins have highest probability models with higher amplitude bursts ( @xmath34 @xmath1 30 ) , relatively long durations ( @xmath32 @xmath1 30 - 40 myr ) , and large periods ( @xmath53 250 myr ) .
the highest probability sfh models along with the corresponding temporal evolution of the h@xmath0-to - fuv ratios are shown in figure [ best_fit_sfhs ] . as a rough proxy for the robustness of the highest probability model
, we rank the models according to the probability of matching the data , as judged by the ks test .
we then examine 10th most probable models in each stellar mass bin and find they have similar parameters to the corresponding highest probability models .
the largest contrast between the best and 10th best models is in the highest mass bin , where the duration increases by a factor of 4 ( from 3 to 12 ) and the amplitude decreases by a factor of 5 ( from 10 to 2 ) .
however , the values of ( @xmath32@xmath33@xmath34)/@xmath35 for the best and 10th highest probability models are comparable ( 0.24 and 0.32 ) .
this simple comparison offers some assurance that we have not simply picked out a single model that by chance describes the observed distribution . along similar lines
, we also consider the 10th lowest probability models in each mass bin .
these models all have vanishingly small probabilities and exhibit parameters that are drastically different from the highest probability models . for example
, the highest probability models favor relatively large bursts in the lowest mass bin , while the 10th worst model is characterized by long periods of constant sf , with bursts of amplitude 2 occurring only @xmath1 3% of the time .
conversely , in higher mass bins where highest probability models are primarily characterized by constant sfhs and modest bursts , the 10th lowest probability models have typical parameters of short duration , high amplitude bursts .
an examination of full parameter space reveals a degree of degeneracy in model parameters .
specifically , models with similar values of ( @xmath32@xmath33@xmath34)/@xmath35 also have comparable ks probabilities .
thus , we consider the highest probability model as a representation of a set of models with similar values of ( @xmath32@xmath33@xmath34)/@xmath35 .
as demonstrated in @xcite , a model sfh that matches the distribution of h@xmath0-to - fuv ratios may not be able to simultaneously account for other observables , such as @xmath2-band surface brightness or stellar mass . with this in mind , in figure [ evol_models ]
, we compare the evolution of the h@xmath0-to - fuv ratios versus stellar mass ( top ) and @xmath2-band surface brightness ( bottom ) over one burst cycle against the observed data .
for illustrative purposes , we have excluded the simulated observational noise from this particular figure .
the relative evolution of stellar mass and surface brightness have been derived entirely from the models .
we centered the starting point of the models on the median values of the observations .
this choice resulted in overlap between points from different models , and in those cases , small adjustments ( @xmath4 0.3 dex ) in the absolute placement of mass or luminosity were made to improve clarity .
loops are composed of discrete model points sampled at time intervals of 0.1 myr . for reference , we have added time labels that correspond to distinct events in the burst cycle to the loop in the lowest surface brightness regime in figure [ evol_models ] ( cyan points ) , which has a period of 250 myr , a burst duration of 40 myr , and an amplitude of 30 ( p250 , d40 , a30 ) .
lccccc @xmath21 @xmath2210.0 & 31.3 & 30.6 & 68.7 & 69.4 + 10.0@xmath24@[email protected] & 35.0 & 36.5 & 65.0 & 63.5 + 9.0@xmath24@[email protected] & 29.1 & 31.5 & 70.9 & 68.5 + 8.0@xmath24@[email protected] & 15.5 & 13.6 & 84.5 & 86.4 + @xmath21 @xmath287.0 & 21.7 & 16.3 & 78.3 & 83.7 [ tab4 ] during a burst , the h@xmath0 and fuv sfrs are not in equilibrium . early in the burst
, the production of h@xmath0 photons exceeds that of the fuv .
however , stars that are fuv bright , are less massive , and have longer typical lifetimes than those responsible for h@xmath0 production .
the difference in stellar lifetimes leads to a build up of fuv emission over the course of the burst , resulting in a net decline in the h@xmath0-to - fuv ratio .
we see this behavior concretely in the magenta ` loop ' in figure [ evol_models ] . immediately following the beginning of a burst ,
( denoted by ` 0 myr ' ) , the h@xmath0-to - fuv ratio is at its peak , as expected . as the burst continues over the subsequent duration of 40 myr
, the h@xmath0-to - fuv ratio steadily declines . following the end of the burst
, we see a significant decrease in the h@xmath0-to - fuv ratio between 40 and 50 myr .
there is negligible h@xmath0 flux after the burst , as the more massive stars expire within a few myr of the termination of star formation .
thus , the build up of fuv emission drives the net h@xmath0-to - fuv ratio to lower values .
in addition , the sfh has returned to a constant , lower level sfr during this 10 myr period , leading to an equilibrium state for the production of h@xmath0 and fuv flux . finally , from 50 to 250 myr , we see a steady increase in the h@xmath0-to - fuv ratio .
the fuv excess , built up during the burst , continues to fade , and the contribution of the constant sfr to the net h@xmath0-to - fuv ratio increases . at 250 myr , near the end of the burst cycle ,
the h@xmath0-to - fuv ratio approaches the equilibrium value .
we now examine the same sequence , only for the evolution of @xmath2-band surface brightness . during the burst , the @xmath2-band surface brightness increases due to the production of luminous young stars , reaching a maximum at the end of the burst . relative to the initial value
, the surface brightness has increased by a value of @xmath31 @xmath1 0.5 .
for the subsequent 10 myr , the @xmath2-band surface brightness decreases .
the death rate of luminous stars exceeds the birth rate , meaning the galaxy can not sustain the maximum surface brightness level obtained during the burst phase , and consequently begins to fade .
finally , from 50 to 250 myr , the @xmath2-band surface brightness continues to fade and approaches the value consistent with that expected from a constant sfh . in figure [ sevol_models ]
, we now consider the model loops including the effects of simulated noise , introduced to mimic uncertainties in the observed fluxes .
the inclusion of noise adds to the dispersion in the models points .
the model points represent equal time intervals of 0.1 myr , such that the density of points is directly proportional to the amount of time spent at a given phase of the burst cycle . from this information ,
we conduct a simple comparison between the fraction of model and data points above and below the fiducial h@xmath0-to - fuv ratio . as shown in table [ tab4 ] , the predictions from the models agree with the observations within @xmath1 5% . in general , @xmath1 30 - 35% of the observed massive galaxy sample is above the fiducial .
the corresponding models predict a similar fraction .
observations of lower mass systems show that @xmath1 15 - 20% of the galaxies are above the fiducial .
the highest probability sfh models in these bins also occupy a similar range , indicating that models of galaxies bursty sfhs naturally account for low mass galaxies with both low and normal h@xmath0-to - fuv ratios . in figure [ sevol_models_ha_mb ]
, we plot the h@xmath0-to - fuv ratios versus sfr(h@xmath0 ) ( top ) and versus @xmath54 ( bottom ) , including the effects of simulated noise as in [ noise ] . in the top panel
, we see that the predicted change in the h@xmath0-to - fuv ratio shows an expected strong correlation with the h@xmath0 based sfr , such that galaxies with higher h@xmath0-to - fuv ratios generally have higher h@xmath0 sfrs . from the observations
, we see that galaxies with low h@xmath0 sfrs and low h@xmath0-to - fuv ratios are quite rare , which agrees with the sfh model predictions . plotted versus @xmath9 , the model sfhs indicate an increase of @xmath1 2 mag in blue luminosity during a burst .
the majority of the time , however , the model sfhs show that galaxies are in lower luminosity states with h@xmath0-to - fuv ratios below the fiducial .
our study is not the first to model the effects of sfhs on h@xmath0-to - fuv ratios . within the past decade three notable studies
@xcite have compared h@xmath0 and fuv fluxes from sets of sfh models with those from observations .
these three studies represented a forward step in considering a range of sfh parameter space beyond past analysis , which were generally restricted to either constant or instantaneous burst models ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
likewise , the analysis presented in this paper provides another incremental step toward understanding the impact of non - constant sfhs on observed h@xmath0-to - fuv ratios .
the primary differences between this study and those of @xcite , @xcite , and @xcite are in the construction and analysis of model sfhs and in the method of correcting for internal dust attenuation . in this section
, we will focus on the differences in model sfhs .
the model sfhs considered in this paper represent a fine sampling of sfh parameter space .
in contrast , two of the three previous studies consider a coarse sampling , generally intended to bracket ranges of sfhs .
for example , @xcite consider 18 single burst and gasp model sfhs with durations of 10 , 10 , 1000 myr and amplitudes of 2 , 10 , and 100 , and find that none are able to simultaneously account for the observed trends of h@xmath0-to - fuv ratios versus h@xmath0 and @xmath2-band surface brightness .
@xcite considered a slightly broader range sfh parameters and analyze 36 model sfhs , but reach an ambiguous conclusion concerning the influence of sfhs on h@xmath0-to - fuv ratios .
our sfh modeling finds that none of the extreme models , i.e. , those considered bracketing the range of parameter space , provide good descriptions of the data .
thus , it is not entirely surprising that these prior studies did not find well matched model sfhs .
lccccc @xmath21 @xmath2210.0 & 31.3 & 30.6 & 68.7 & 69.4 + 10.0@xmath24@[email protected] & 35.0 & 36.5 & 65.0 & 63.5 + 9.0@xmath24@[email protected] & 29.1 & 31.5 & 70.9 & 68.5 + 8.0@xmath24@[email protected] & 15.5 & 13.6 & 84.5 & 86.4 + @xmath21 @xmath287.0 & 21.7 & 16.3 & 78.3 & 83.7 [ tab4 ] the ` micro - sfhs ' considered by @xcite represent a limited , but fine sampling of sfh parameter space .
the focus of micro - sfhs is on the effects of the more recent burst on the h@xmath0 luminosity .
episodic bursts with durations of 2 , 5 , and 10 myr over the course of 100 myr were considered to simulate the effects of various hii regions turning on and off .
the principle behind this method is similar to our analysis , except we consider a larger number of possible model permutations and the contribution of ancient stellar populations , allowing us to accurately assess the impact of recent sfhs on stellar mass and surface brightness , as well as the observed h@xmath0 and fuv fluxes .
we further note that the models considered by @xcite are similar to those in our highest mass bins , but they do not consider models with parameters such as those in our lowest mass bins .
this seems appropriate , however , as the sample of @xcite does not extend to comparably low luminosity galaxies .
a further point of contrast between our study and the previous three is the division of the sample for analysis .
each of the previous studies attempt to find a single sfh that best describes the entire observed sample . by binning the data in increments of stellar mass
, we do not impose the restriction that a single model sfh must be able to explain the observed h@xmath0-to - fuv ratios in both massive galaxies such as m51 and low mass dwarfs such as gr8 .
indeed , as we have shown , the highest probability models converge on drastically different characteristic sfhs for galaxies of different masses .
an underlying assumption to any calibration of sfr indicators is the nature of the imf . in the conversion from observed flux to sfr
, the imf is assumed to be universal , i.e. , the same with respect to time and environment , and fully populated , i.e. , the relative number of stars per unit mass interval is always the same .
however , whether either of these criteria are true in nature remains a heavily debated and open question ( see the review by @xcite ) .
unfortunately , only a handful of variable imf scenarios in the literature have been discussed in the context of h@xmath0-to - fuv ratios . extracting results from some of these studies ,
we proceed to discuss these findings alongside predictions from the model sfhs analyzed in this paper .
we first consider the case of a universal , but not fully sampled imf .
the effect of this stochastically sampled imf scenario on the h@xmath0-to - fuv ratio has been simulated in @xcite . assuming a constant sfh , the authors found that the predicted turnover in h@xmath0-to - fuv ratios occurred at too low an h@xmath0 luminosity relative to the data ( see figure 7 in @xcite ) .
that is , randomly sampling a universal imf does not appear to adequately account for the observations .
this result is confirmed by more recent simulations in @xcite , @xcite , and @xcite . in each paper , the authors verify the inability of random sampling of the imf to account for the observed trend in h@xmath0-to - fuv ratios , under the assumption of a constant sfh and a fully populated cluster mass function .
instead , each study demonstrates that stochastic sampling of _ both _ the stellar and cluster mass functions are necessary to produce h@xmath0-to - fuv ratios that are consistent with observations .
the effect of stochastically sampling the cluster mass function is that lower mass galaxies form massive clusters less frequently than higher mass systems . along the same lines
, stochastically sampling stellar imf implies that high mass stars have a higher probability of forming in high mass clusters . in light of these scenarios
, we would then expect low mass galaxies to form fewer high mass clusters , and hence fewer high mass stars when compared to more massive galaxies .
therefore , low mass systems would have fewer massive stars capable of producing h@xmath0 , and the h@xmath0-to - fuv ratios in low mass systems is expected to be less than for more massive galaxies .
crucially , these models posit that the formation of massive clusters in low mass galaxies are rare , but not impossible , which is the case for certain prescriptions of a variable imf ( see [ igimf ] ) . more detailed discussion and quantitative analysis of these effects can be found in @xcite , @xcite , and @xcite .
the second possibility we consider is that the imf is not universal .
this scenario has been widely debated in the literature , but a consensus has yet to be reached ( e.g. , * ? ? ?
we focus on one particular scenario , the integrated galactic imf ( igimf ) as detailed in @xcite , which makes specific predictions for h@xmath0-to - fuv ratios as functions of h@xmath0 luminosity @xcite .
there are three tenets of the igimf : @xmath55 all stars are formed in clusters ; @xmath56 the maximum mass of a cluster star is determined by the mass of the cluster in which it is formed ; @xmath57 the maximum cluster mass is a deterministic function of the integrated sfr of a galaxy . from these three postulates ,
the igimf predicts a systematic increase in the ratio of h@xmath0-to - fuv fluxes as a function of h@xmath0 luminosity , under the assumption of a constant underlying sfh @xcite . in what follows
, we will compare predictions from the igimf and model sfhs with our observations .
first , we will consider the standard case of the h@xmath0-to - fuv ratios as a function of h@xmath0 luminosity .
the igimf predictions in this case are taken directly from the work of @xcite .
we will then compare the model predictions to the data as a function of galaxy stellar mass . in this case , to compute the galaxy stellar mass for the igimf models , we have assumed a constant sfh over the history of the galaxy at a level of the current h@xmath0 sfr .
we discuss on the validity of this assumption below . as a first comparison ,
we follow the discussion of the igimf in @xcite , and over - plot predictions for the ` standard ' and ` minimum1 ' igimf models ( solid and dashed navy lines ) on the observed h@xmath0-to - fuv ratios versus h@xmath0 luminosity in the top panel of figure [ igimf_comp ] .
the two igimf models correspond to different convolutions of the cluster mass function with a stellar imf , such that the minimal1 model is the least deviant from a standard salpeter imf ( see @xcite for specific details ) .
we have adopted the igimf models from @xcite that assume a constant sfh over the history of the universe .
we have also over - plotted the predictions from the highest probability model sfhs .
each of the colored points represents the median h@xmath0-to - fuv ratio from models shown in figure [ best_fit_sfhs ] and listed in table [ tab3 ] .
the placement on the horizontal axis corresponds to the median h@xmath0 luminosity in each mass bin for the observed sample .
the error bars represent the 5th and 95th percentiles ( @xmath1 2-@xmath12 ) in the model distributions . in the top panel of figure [ igimf_comp ]
, we see that both the igimf and sfh models qualitatively agree with observations . at high h@xmath0 luminosities ,
both types of model indicate that ratios should be near the fiducial . for decreasing h@xmath0 luminosity ,
both models predict a decline in the h@xmath0-to - fuv ratio that is in reasonable agreement with the mean observational trends .
of the two igimf scenarios , the ` minimal1 ' model appears to more closely follow the mean of the data better than the ` standard1 ' model , particularly when the data are considered as a function of stellar mass .
the scatter in the h@xmath0-to - fuv ratios is another quantity we consider . for h@xmath0 luminosities
@xmath7 10@xmath58 , we see a number of data points that lie within the range of the sfh model predictions , but above the two igimf curves .
we note that it is not clear how the igimf can explain galaxies that lie significantly above the predicted curves , and pursue a more detailed discussion of this point below . in the bottom panel of figure [ igimf_comp ] ,
we repeat the comparison outlined above , only this time we plot galaxy stellar mass on the x - axis .
again , we see that both the sfh and igimf models predict a general decline in the h@xmath0-to - fuv ratios as a function of decreasing stellar mass
. however , both igimf models tend to under predict the h@xmath0-to - fuv ratios for galaxies @xmath4 10@xmath59 m@xmath6 . in comparison
, the median values of the sfh models do not decrease as sharply as those of the igimf , and instead exhibit a larger dynamic range of possible model values . of particular interest
is the relationship between the two types of models and the scatter in the data .
the broad dispersion of h@xmath0-to - fuv ratios predicted by the model sfhs are generally in good agreement with the observations .
the most notable discrepancies are the handful of galaxies that have h@xmath0-to - fuv ratios that are lower than predicted by the models .
this may be an indication that stochastic imf effects need to be included to explain the extremely low h@xmath0-to - fuv points .
the igimf models are generally not in good agreement with the observed scatter . in the best case ,
the igimf models track the decline of average h@xmath0-to - fuv ratios as a function of decreasing h@xmath0 luminosity , but there are a number of galaxies with ratios significantly below the igimf predicted curves , and a smaller fraction that lie above . the disagreement worsens when the comparison is done as a function of stellar mass .
in particular , for galaxies with stellar masses @xmath4 10@xmath59 m@xmath6 , there are a substantial number of galaxies that have h@xmath0-to - fuv ratios significantly above the igimf predictions .
these data points are challenging to understand in the framework of the igimf .
taken at face value , the igimf is a deterministic theory ( e.g. , * ? ? ?
that is , it makes specific predictions about the the functional form of the h@xmath0-to - fuv ratio , namely that h@xmath0 should be deficient in galaxies with low masses or low sfrs . yet
, the data show a number of low mass systems with normal or high h@xmath0-to - fuv ratios , implying the frequent presence of massive star(s ) in low mass systems ; a observational finding that does not seem compatible with the tenets of the igimf
. admittedly , there are two areas of uncertainty with this interpretation .
first , the connection between total galaxy stellar mass and the igimf is not well defined in the literature .
although such a connection has been invoked in other analyses of the igimf ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , extrapolating the current h@xmath0 sfr to a constant lifetime sfh to determine the stellar mass is not a valid assumption . in the event that we have _ over - estimated _ the galaxy stellar mass , the igimf curves would shift to the left in the bottom panel of figure [ igimf_comp ] . in order to match
the mean observed flux ratios , the extrapolated galaxy masses need to be decreased by a factor of @xmath1 10 .
even with this correction applied , the igimf curves still do not match the observed scatter in the data very well .
the second area of uncertainty is the connection between the igimf and assumed form of the sfh .
both the ` standard ' and ` minimal1 ' igimf models assume a constant sfh ( e.g. , * ? ? ?
* ; * ? ? ?
including the effects of a bursty sfh may be able to produce h@xmath0-to - fuv ratios higher than currently predicted . as seen from the model sfhs , during the early phases of a burst , the h@xmath0-to - fuv ratios are near their peak .
combining this mechanism with the igimf provides at least one possible explanation for galaxies with higher than predicted h@xmath0-to - fuv ratios
. however , the convolution of bursts with the igimf is not without complications .
first , if bursts are required to explain h@xmath0-to - fuv ratios above the igimf predictions , this implies that all such galaxies are currently in the midst of a burst .
however , it seems unlikely that the majority of low mass galaxies in our sample should be undergoing simultaneous bursts .
second , under the auspices of the igimf , one also needs to consider the absolute sfr , not just the relative burst amplitudes . at low absolute sfrs such as 10@xmath60 m@xmath6 yr@xmath61
, the igimf model predict that @xmath42 $ ] is @xmath1 @xmath101 . extrapolating this sfr over 13.7
gyr yields a stellar mass of @xmath1 10@xmath62 m@xmath6 . at this stellar mass ,
we see a handful of galaxies with @xmath42 $ ] @xmath7 0 .
the only way for such low mass galaxies to have these high h@xmath0-to - fuv ratios is through intense bursts of sf that significantly increase the amount of h@xmath0 emission .
additionally , we note that such bursts should not be too long in duration or too high in frequency or the final mass of the galaxy could be substantially increased .
overall , we find that bursty sfh models can qualitatively account for the general decline and scatter in the observations as a function of decreasing h@xmath0 luminosity or galaxy stellar mass , whereas the simplest form of the igimf models do not .
we emphasize that this finding does not invalidate the igimf models , nor suggest that a variable imf is not possible , only that there appear to be other viable explanations for the observed h@xmath0-to - fuv ratios that do not involve a systematically varying imf . indeed , we believe that future model efforts that include more complex sfhs , and reasonable variations on the imf are essential for understanding sf processes in galaxies .
we suggest some possible paths forward in the following section . in this paper
, we have found a set of sfh models that match the trend and scatter in observed h@xmath0-to fuv ratios .
however , we caution that the broad applicability of our conclusions should be tempered by the inherent simplicity of the models .
sfhs in nature are undoubtably more complex than periodic bursts with fixed amplitudes and durations .
we have considered characteristic sfh models for galaxies in comparable mass ranges , but it is likely that even galaxies with similar stellar masses do not share _ identical _ sfhs .
indeed , studies within the local group and nearby universe have shown that galaxies with similar masses exhibit diversity in their sfhs ( e.g. , * ? ? ?
* ; * ? ? ?
* ) . determining the degree to which any model sfh truly reflects those in nature is a challenging and presently unresolved question .
a comprehensive assessment of the realistic nature of the model sfhs used in this paper is beyond the scope of this study , but is the subject of future work ( e.g. , * ? ? ?
* ; * ? ? ?
instead , in this section we outline the limitations of the simple model sfhs and discuss ways in which future sfh model efforts can be improved .
one increasingly common method of directly measuring sfhs is from analysis of color - magnitude diagrams ( cmds ) of resolved stellar populations in nearby galaxies ( see @xcite and references therein ) . in figure [ real_sfhs ] , we show an example of six such sfhs measured from cmds based on hubble space telescope imaging of dwarf galaxies in the m81 group @xcite .
the solid black and dot - dashed magenta lines represent the same sfh binned to different time resolutions , which we discuss below .
these measured sfhs provide an indication of the complexity of galaxy - wide star formation .
the most striking characteristic in these measured sfhs is the degree of stochasticity in the amplitude and duration of star formation episodes .
further , there is no clearly defined period associated with any of the measured sfhs .
there is also a conspicuous lack of stochasticity in the model sfhs ( e.g. , figure [ sample_sfhs ] ) .
clearly , non - uniform parameterization of bursty sfhs is an important next step toward constructing more realistic sfhs .
of course , there is some degree of similarity between the modeled and measured sfhs .
focusing on the most recent 100 myr ( insets in figure [ real_sfhs ] ) , where the time resolution of the cmd - based sfhs is the highest , a handful of bursts with amplitudes of @xmath1 10 - 50 are present in the measured sfhs , similar to those found in the highest probability models for low mass galaxies .
the measured sfhs also suggest that extended duration , high amplitude bursts ( e.g. , @xmath63 100 myr , @xmath64 50 ) are probably not common . burst durations of hundreds of myrs have been measured in a minority of starburst dwarf galaxies @xcite , but with average burst amplitudes of less than a factor of 5
. however , these global , long duration bursts are superimposed with a short timescale ` flickering ' of enhanced sf , which result in short - term burst amplitudes @xmath7 10 on 10 myr timescales .
average burst amplitudes in these systems are typically a factor of a @xmath4 5 over the course of hundreds of myr , although variations in the sfrs on timescales of order @xmath1 10 myr can vary by factors of 10 or more
. determinations of exact amplitude and duration parameters directly from cmd - based sfhs are challenging due to the logarithmic time resolution of the measured sfhs .
we explicitly illustrate the issue at hand in appendix [ timeres ] .
integrated fluxes provide further empirical constraints for sfh modeling efforts
. a realistic set of model sfhs should be able to reproduce such observations as the distribution of observed h@xmath0 ews , the uv luminosity function , and optical broadband colors and luminosities .
the simple model sfhs we consider are not in strong agreement with either observed distributions of either h@xmath0 ew or uv luminosity , as presented in @xcite and @xcite . while both the model and observed distributions are well described by lognormal functions , the functional form of the model predictions has too many galaxies near the median value and not enough at the extremes . some of the difference can be reconciled by increasing the simulated noise in the models , but we would prefer to avoid this type of fine - tuning .
further , the addition of stochasticity would also mitigate the over - density of galaxies near the median of the distribution .
quantitatively testing these effects is beyond the scope of this paper , but is discussed in the context of the slug population synthesis code @xcite .
we have compared predictions from simple , periodic sfhs with observed h@xmath0-to - fuv sfr ratios observed as part of the 11hugs and _ spitzer _ lvl programs . from a suite of @xmath1 1500 simple model sfhs
, we have identified a set of models that matches the observed distribution of h@xmath0-to - fuv sfr ratios over a range of galaxy stellar masses .
we find that high mass galaxies tend to have characteristic sfhs that are predominantly constant , with relatively modest amplitude bursts ( @xmath34 @xmath4 10 ) spaced by @xmath1 100 myr .
we find that lower mass galaxies are best described by sfhs with burst amplitudes of @xmath1 30 , and inter - burst spacings of @xmath1 150 - 200 myr .
these model sfhs have burst amplitudes that are comparable to those inferred from cmds of nearby star forming dwarf galaxies . using results from the highest probability models , we compared the predicted evolution of the h@xmath0-to - fuv ratios versus stellar mass , @xmath2-band surface brightness , sfr(h@xmath0 ) , and @xmath9 , and found that in all cases , the models were in good agreement with observations .
we find that highest probability models are well matched to the observed systematic decline in the h@xmath0-to - fuv ratios as a function of both decreasing h@xmath0 luminosity and decreasing galaxy stellar mass .
variations in the high mass stellar imf have been proposed as an explanation for this trend .
we therefore compare predictions from the model sfhs with those from the igimf theory , which also makes specific predictions for h@xmath0-to - fuv ratios . from this comparison
we see that stochastic sampling of the imf can not re - produce the observed trend .
a convolution of stochastically sampled stellar and cluster mass functions are necessary to match observations .
we find that both the sfh and igimf predictions are in good agreement with the h@xmath0-to - fuv ratios as a function of h@xmath0 luminosity .
however , when considered versus stellar mass , it is not clear if the igimf allow for h@xmath0-to - fuv ratios higher than the predicted curves , whereas the model sfhs naturally account for the observed scatter . instead ,
the we find that only time variable sfhs are able to explain both the mean trend in the data , as well as the scatter .
while we have only considered toy model sfhs , they share characteristics with the stochastically sampled cluster and stellar mass function scenario presented in @xcite , @xcite , and @xcite .
essentially , at low sfrs , a stochastically sampled cluster mass function rarely produces massive clusters .
however , when a massive cluster does form , the sfr of the galaxy increases , and can effectively look like a burst of sfh . while the frequency of massive cluster formation in low mass galaxy is not well understood , the new population synthesis code , slug @xcite will permit investigations on the impact of stochastic cluster formation on the galaxy wide sfh and broadband flux properties .
finally , we discuss the limitations of our toy model sfhs , including the assumptions of periodicity and lack of stochasticity . in addition
, we find that one major challenge is understanding differences between our model predictions and observations of the h@xmath0 and uv luminosity functions , as well as precisely matching observed optical and nir galaxy luminosities . exploring these differences are one of the topics that will be explored in the upcoming paper of @xcite .
the authors would like to thank the anonymous referee for useful suggestions that helped to improve this paper .
the authors would also like to thank nate bastian , john eldridge , and robert da silva for insightful discussions about the nature of the imf .
drw is grateful for support from the university of minnesota doctoral dissertation fellowship and penrose fellowship .
this work is based on observations made with the nasa / esa hubble space telescope , obtained from the data archive at the space telescope science institute .
support for this work was provided by nasa through grants go-10605 and go-10915 from the space telescope science institute , which is operated by aura , inc .
, under nasa contract nas5 - 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 extensive use of nasa s astrophysics data system bibliographic services .
widest - label abazajian , k. n. , et al .
2009 , , 182 , 543 annibali , f. , greggio , l. , tosi , m. , aloisi , a. , & leitherer , c. 2003 , , 126 , 2752 bastian , n. , covey , k. r. , & meyer , m. r. 2010 , , 48 , 339 bell , e. f. , & de jong , r. s. 2001 , , 550 , 212 bell , e. f. , & kennicutt , r. c. , jr . 2001 , , 548 , 681 blanton , m. r. , et al .
2005 , , 129 , 2562 boselli , a. , boissier , s. , cortese , l. , buat , v. , hughes , t. m. , & gavazzi , g. 2009 , , 706 , 1527 bressan , a. , fagotto , f. , bertelli , g. , & chiosi , c. 1993 , , 100 , 647 bruzual , g. , & charlot , s. 2003 , , 344 , 1000 bruzual , g. 2007 , from stars to galaxies : building the pieces to build up the universe , 374 , 303 buat , v. , et al .
2005 , , 619 , l51 buat , v. 1992 , , 264 , 444 buat , v. , donas , j. , & deharveng , j. m. 1987 , , 185 , 33 cardelli , j. a. , clayton , g. c. , & mathis , j. s. 1989 , , 345 , 245 chabrier , g. 2003 , , 115 , 763 da silva , r. l. , fumagalli , m. , & krumholz , m. 2011 , arxiv:1106.3072 dalcanton , j. j. , et al .
2009 , , 183 , 67 dale , d. a. , et al .
2009 , , 703 , 517 dale , d. a. , & helou , g. 2002 , , 576 , 159 dolphin , a. e. , weisz , d. r. , skillman , e. d. , & holtzman , j. a. 2005 , arxiv : astro - ph/0506430 eldridge , j. j. 2011 , arxiv:1106.4311 elmegreen , b. g. 2006 , , 648 , 572 fagotto , f. , bressan , a. , bertelli , g. , & chiosi , c. 1994 , , 105 , 29 fumagalli , m. , da silva , r. l. , & krumholz , m. r. 2011 , arxiv:1105.6101 glazebrook , k. , blake , c. , economou , f. , lilly , s. , & colless , m. 1999 , , 306 , 843 hao , c .-
n . , kennicutt , r. c. , jr . , et al .
2011 , 2011 , , accepted iglesias - pramo , j. , boselli , a. , gavazzi , g. , & zaccardo , a. 2004 , , 421 , 887 johnson , b. , et al .
, in prep .
johnson , l. c. , et al . , in prep .
kennicutt , r. c. , et al .
2009 , , 703 , 1672 kennicutt , r. c. , jr .
, lee , j. c. , funes , s. j. , jos g. , sakai , s. , & akiyama , s. 2008 , , 178 , 247 kennicutt , r. c. , jr .
1998 , , 36 , 189 lee , j. c. , kennicutt , r. c. , jos g. funes , s. j. , sakai , s. , & akiyama , s. 2009 , , 692 , 1305 lee , j. c. , kennicutt , r. c. , jos g. funes , s. j. , sakai , s. , & akiyama , s. 2009 , , 692 , 1305 lee , j. c. , et al .
2009 , , 706 , 599 lee , j. c. , et al . 2011 , , 192 , 6 leroy , a. k. , walter , f. , brinks , e. , et al .
2008 , , 136 , 2782 mcquinn , k. b. w. , et al .
2010 , , 721 , 297 mcquinn , k. b. w. , et al .
2010 , , 724 , 49 meurer , g. r. , et al .
2006 , , 165 , 307 meurer , g. r. , et al .
2009 , , 695 , 765 moorwood , a. f. m. , van der werf , p. p. , cuby , j. g. , & oliva , e. 2000 , , 362 , 9 pflamm - altenburg , j. , weidner , c. , & kroupa , p. 2009
, , 395 , 394 pflamm - altenburg , j. , weidner , c. , & kroupa , p. 2007
, , 671 , 1550 salim , s. , et al .
2007 , , 173 , 267 schlegel , d. j. , finkbeiner , d. p. , & davis , m. 1998 , , 500 , 525 sullivan , m. , treyer , m. a. , ellis , r. s. , & mobasher , b. 2004 , , 350 , 21 sullivan , m. , treyer , m. a. , ellis , r. s. , bridges , t. j. , milliard , b. , & donas , j. 2000 , , 312 , 442 tolstoy , e. , hill , v. , & tosi , m. 2009 , , 47 , 371 van zee , l. , et al . , in prep .
weidner , c. , & kroupa , p. 2005
, , 625 , 754 weisz , d. r. , skillman , e. d. , cannon , j. m. , dolphin , a. e. , kennicutt , r. c. , jr . , lee , j. , & walter , f. 2008 , , 689 , 160 weisz , d. r. , et al .
2011 , arxiv:1101.1093 weisz , d. r. , et al . , in prep .
yan , l. , mccarthy , p. j. , freudling , w. , teplitz , h. i. , malumuth , e. m. , weymann , r. j. , & malkan , m. a. 1999 , , 519 , l47
in principle , one can synthesize h@xmath0-to - fuv ratios from cmd based sfhs , eliminating the need for the intermediate step of modeling sfhs .
however , the non - uniformly linear time resolution of cmd based sfhs presents a challenge to such efforts .
much like stellar isochrones , cmd based sfhs are naturally sampled in uniform logarithmic time steps , which are not equally spaced in linear time .
the combined coarseness of the time binning for lookback times @xmath7 100 myr and the sensitivity of the h@xmath0-to - fuv ratios to the input parameters limit the utility of cmd based sfhs for this particular purpose . as a simple demonstration of the effects of time resolution , in figure [ real_sfhs ]
, we show the same sfhs binned at the highest time resolution ( solid black ) and the resolution adopted by @xcite ( dot - dashed magenta ) .
although these represent the same underlying sfh , the choice in time - binning gives a strikingly different impression of the galaxy sfhs . extending this exercise to the models considered in this paper , in figure [ res_sfhs ] we plot the example model sfhs from figure [ sample_sfhs ] convolved to the two cmd based time binning schemes . in most cases , the input sfhs are sufficiently degraded to an unrecognizable state .
there is , of course , some dependence on the particular sfh , as well as the quality of the observed cmd . a more detailed discussion of time resolution and sfh recovery from cmd can be found in @xcite . | we consider the effects of non - constant star formation histories ( sfhs ) on h@xmath0 and galex far ultra - violet ( fuv ) star formation rate ( sfr ) indicators . under the assumption of a fully populated chabrier imf
, we compare the distribution of h@xmath0-to - fuv flux ratios from @xmath1 1500 simple , periodic model sfhs with observations of 185 galaxies from the _ spitzer _ local volume legacy survey .
we find a set of sfh models that are well matched to the data , such that more massive galaxies are best characterized by nearly constant sfhs , while low mass systems experience bursts amplitudes of @xmath1 30 ( i.e. , an increase in the sfr by a factor of 30 over the sfr during the inter - burst period ) , burst durations of tens of myr , and periods of @xmath1 250 myr ; these sfhs are broadly consistent with the increased stochastic star formation expected in systems with lower sfrs .
we analyze the predicted temporal evolution of galaxy stellar mass , @xmath2-band surface brightness , h@xmath0-derived sfr , and blue luminosity , and find that they provide a reasonable match to observed flux distributions .
we find that our model sfhs are generally able to reproduce both the observed systematic decline and increased scatter in h@xmath0-to - fuv ratios toward low mass systems , without invoking other physical mechanisms .
we also compare our predictions with those from the integrated galactic imf theory with a constant sfr .
we find that while both predict a systematic decline in the observed ratios , only the time variable sfh models are capable of producing the observed population of low mass galaxies ( @xmath3 @xmath4 10@xmath5 m@xmath6 ) with normal h@xmath0-to - fuv ratios .
these results demonstrate that a variable imf alone has difficulty explaining the observed scatter in the h@xmath0-to - fuv ratios .
we conclude by considering the limitations of the model sfhs , and discuss the use of additional empirical constraints to improve future sfh modeling efforts . |
one of the most fundamental processes in light - matter interaction is the attenuation of light by matter .
the introduced weakening of light is observed since ancient times . today , in modern nano - optics , it is possible to perform controlled experiments with single entities , i.e. single photons @xcite and single emitters @xcite .
the theoretical description of this optical interaction does not only address very fundamental questions , it also allows the optimization of the efficiency in light - matter interaction .
the spectroscopy and optical research on single emitters in the past was engineered for optimal light - extraction strategies @xcite , such that e.g. a single molecule or a nano - crystal can be detected by fluorescence with a good signal to noise ratio .
but it was rarely optimized to achieve the most efficient excitation . for optical protocols , such as the quantum phase gate , or coherent microscopy schemes @xcite ,
an efficient interaction of a single photon to an external field , ideally a single photon , is desirable .
experimentally , single molecules @xcite , quantum dots @xcite , and atoms @xcite were researched in the past decade for their ability to attenuate a light beam .
the solid state experiments were initially conducted under cryogenic conditions @xcite .
only balanced detection made it possible to extend these fundamental studies to room temperature @xcite .
motivated by the experimental realization of these fundamental processes , numerous theoretical approaches were reinvented @xcite , often founded on electrodynamic calculations from the 1960s when the field of antenna theory was very active .
one central question was the maximum amount of extinction in a real optical focus @xcite .
others describe the energy relations in the context of the optical theorem and in the near - field @xcite .
the energy flow itself , depicted as streamlines of the poynting vector , introduced a vivid insight in this fundamental problem already decades ago @xcite .
today , we see the most fundamental approaches are getting unified between classical antenna theory , quantum technologies and nano - optics .
in this paper we review the fundamental basics of coherent light - matter interaction and absorption of a single emitter .
the relevant cross sections of absorption , extinction and scattering are analytically derived from first principles .
everything is extended to the mathematical description of the energy flow in the proximity of a dipolar emitter .
while this was discussed already in the past decade , we address shortcomings and present novel findings .
the cross sections are calculated , and the analytic result and their derivation by following the energy flow are compared . in the close proximity of the dipole the total field
is altered by the presence of the emitter .
this change also alters the effective cross sections .
the ( optical ) near - field cross sections become more and more a dipolar shape . whereas the dipolar emitter was assumed as a linear , hertzian , dipole in many papers , we see the experimental efforts extended to single atoms @xcite , which usually exhibit a circular dipole .
therefore , we adapt the underlying math to the case of a circular dipole .
we approach the relevant math with an intuitive picture of the fundamental process of light - matter interaction . to that end
we study the most basic case , given by the interaction of an incident plane wave with a single dipole .
a single dipole emitter is assumed in free space at the origin of a coordinate system ( fig .
[ fig:01]a ) . an incident plane wave , propagating from @xmath0 to @xmath1 excites the emitter .
this field is expressed as @xmath2 if we consider a electrically polarizable system such as a dipole in vacuum , excited by a incident electric field the response of the system due to the incident field is given by @xmath3 where @xmath4 is the electric field at the position of the emitter , @xmath5 is the electric polarizability , and @xmath6 is the induced dipole moment .
we consider a parallel response of the system due to the incident field .
the system is treated as a point - like dipole , i.e. infinitely small . in general , the response of the system is not necessarily parallel to the incident field , and @xmath5 has to be written as a tensor . as a consequence of the excitation , the dipole scatters a certain amount of power into all @xmath7 steradians of the environment ( fig .
[ fig:01]a ) . in the following
we calculate the time averaged energy flow into the system , to get an idea how energy is absorbed by a dipole .
to do so , we calculate the time averaged poynting vector , computed by ( for time harmonic fields ) @xcite @xmath8 where @xmath9 is the total electric field calculated as a superposition of the incident field and the scattered field and @xmath10 is the corresponding magnetic field . while the incident field is given by , red ) and backward ( @xmath11,blue ) component dependent on the radius of the sphere as well as @xmath12 .
the situation corresponding to d ) is marked with a gray dashed line .
d ) solutions for @xmath13 ( blue , dashed ) and @xmath14 ( green ) .
the reference curve @xmath15 defines zero .
the magnitude of the solutions is plotted against the reference curve . ]
@xmath16 and the scattered dipole field is expressed by @xcite @xmath17\left(\frac{1}{r^3}-\frac{\mathrm i k}{r^2}\right ) \right\ } \exp{(\mathrm i k r ) } \\ \vec{b}_{\mathrm{scat}}&= & k^2 d_0 ( \hat r\times \hat{d})\left(1-\frac{1}{\mathrm i k r}\right ) \frac{\exp{(\mathrm i k r)}}{r}\;.\end{aligned}\ ] ] where @xmath18 is the unit vector in the direction of the incident polarizability , @xmath19 is the unit vector in the direction of the induced dipole moment and @xmath20 is the magnitude of the dipole moment ( @xmath21 ) . since we said that the dipole moment is parallel to the incident wave
, @xmath22 is valid .
the total energy flow is written as @xmath23 if one compares the situation we investigate with ref .
@xcite , it is important to notice that we have no gouy phase shift of 90@xmath24 , which is only present in an optical focus . as a consequence , we have no equivalent effect of back reflection as introduced there .
this additional phase would lead to a `` perfect reflection '' in the backward direction , respectively to full attenuation in the forward direction @xcite .
we now separate the space into two half spheres , one in the forward direction and one in the backward direction . as shown in fig .
[ fig:01]a , we find that the sum of the energy contributions traveling through the sphere in the forward direction ( 2 ) and in the backward direction ( 1 ) yield to the same amount with different signs so that their total sum is zero as required by energy conservation . the different energy contributions and their signs , i.e. if they are directed inwards to ( - ) or outwards from the sphere ( + ) , are summarized in the table fig .
[ fig:01]b . in this listing
we consider the different parts of the total energy flow .
the extinction terms are here listed as negative , since the extinguished power is commonly defined by @xmath25 ( blue , dashed ) and @xmath26 ( green ) for values of @xmath27 chosen in distances of @xmath28 .
one anti - node occurs within half a wavelength .
this periodic behavior has been described in context of fig .
[ fig:01]c .
the plots shown here are closely related to the results of ref .
@xcite . ]
therefore , the negative signs in the list shown in fig . [ fig:01]b indicate , that the cross terms yield to an energy flow towards the dipole in the forward direction ( 2 ) .
the contributions of the incident wave are obvious and equivalent ( resp . with different signs ) for the two half spheres .
the scattered energy is greater than zero and has the same amount in both directions i.e. the same amount of power is scattered in the forward and backward direction as expected ( dipole radiation ) . the contributions which may not be directly intuitive are the extinction terms .
the results of the integration over @xmath29 is shown in fig .
. it might be confusing that this integration is not independent of @xmath27 ( radius of the sphere ) .
the incident plane wave has no constant phase on the surface of a half sphere , thus the integral depends on the radius and therefore on how many wave peaks and valleys are collected .
this is the reason that fig .
[ fig:01]c shows a periodic behavior in @xmath30 .
moreover , fig .
[ fig:01]c shows , that the sum of the integration over the extinction terms yields always to @xmath31 as claimed . neglecting the oscillations caused by the incident wave
it becomes clear that extinction occurs only in the forward direction and the amount adds up to @xmath32 ( c.f .
[ fig:03]c ) , which corresponds to the definition of extinction as scattering plus absorption @xcite , since we have no absorption ( such as a transfer to heat ) .
[ fig:01]d shows the extinction part of the poynting vector in the @xmath33- as well as in the @xmath34-plane .
we have a small constant effect on the @xmath35-axis in the forward direction , independent of the radius ( c.f .
[ fig:02 ] ) .
this can be seen as a small deviation of the reference level @xmath15 and @xmath36 .
this effect occurs , since the phase relation between the scattered and incident wave is constant on the @xmath35-axis .
if the behavior of this plot is studied dependent on the radius ( @xmath27 ) , one anti - node occurs within a half wavelength as shown in fig .
[ fig:02 ] .
moreover , there is a phase shift of @xmath37 between the @xmath38 ( blue , dashed ) and the @xmath39 ( green ) solutions on the incident half - sphere ( 1 ) .
the sum of the two parts in the forward and the backward direction always yield to zero as expected by energy conservation . since extinction is given by the sum of absorption and scattering it is computed by the total amount of energy extracted from the incident field by the emitter .
we consider the time averaged power extracted by the dipole by @xcite @xmath40 here , @xmath41 is the electric current density ( a harmonic time dependence @xmath42 was used ) . for eqn .
[ formel3 ] it is essential that the electric field is uniform and time harmonic over the volume , @xmath43 , of the system ( semi - classical approach ) .
the time averaged energy of the incident wave is @xmath44 an oscillating dipole radiates power and the the scattered energy is given by @xcite @xmath45 if no energy is transferred into other channels ( e.g. heat ) the amount of extracted and scattered energy has to be equal .
this implies that the absorption is zero . in the time
averaged case no energy flows into the dipole .
all energy which is transferred to the emitter first excites the emitter and is then re - radiated .
we now introduce the situation where the emitter is coupled to another bath that can dissipate some of the absorbed power from the incident field .
this diminishes the re - radiation of energy and the oscillator is a sink for the energy .
the loss contribution of the polarizability can be divided into two parts : one due to scattering loss , described by @xcite @xmath47 and one due to absorption of energy , i.e. additional decay of the system , e.g. into heat .
the total inverse polarizability can be generally written as @xmath48 in this notation @xmath49 corresponds to a resonant excitation and @xmath50 implies no polarizability at all .
the last term in eqn .
[ formel7 ] corresponds to the scattering loss factor and @xmath51 refers to the absorption coefficient of the system .
for @xmath52 we have a loss - less system
i.e. the whole extracted energy is re - radiated by the dipole as introduced before . in the following ,
we want to have a more detailed look at the resonant case ( @xmath49 ) .
the ideal case , i.e. no absorption and under resonant excitation , should be used as a reference .
therefore , we write the polarizability as @xmath53 this allows the introduction of a loss parameter , @xmath46 , which is sometimes also introduced as a `` single - scattering albedo '' , @xmath54 the dipole moment can be written as @xmath55 in this notation the factor @xmath46 can be seen as a heuristically introduced absorption parameter , since the polarizability of the system is reduced if absorption is introduced .
the different energy contributions on resonance are given by @xmath56 where the relation @xmath57 was used .
the three different contributions in eqn .
[ formel19 ] correspond to the different energy parts given in eqns .
[ formel11],b , c via an integration over an imaginary sphere @xmath58 ( c.f .
[ fig:01 ] ) which includes the dipole .
+ since @xmath46 is given by the ratio @xmath59 we see that for @xmath60 the amount of absorbed power dominates the ratio , and for @xmath61 the amount of scattered energy dominates ( c.f . fig .
[ fig:03]a ) .
equation [ formel12 ] shows the meaning of @xmath46 as the fraction of re - radiated energy to the total extracted energy . in this context
it should be mentioned that g. wrigge derives an analogous result for a two - level - system with decay channels ( below saturation ) in @xcite , that shows the universality of this fact illustrated in fig .
. it might be confusing that some textbooks ( e.g. @xcite ) handle eqn .
[ formel3 ] as the _ absorbed _ power .
if they do so , they use a quasi - static description of the polarizability . in this case scattering by the system is neglected , therefore extinction and absorption become the same .
it has to be mentioned that the quasi static polarizability conflicts with the optical theorem , but eqn .
[ formel7 ] provides a solution to this dilemma . .
at @xmath62 the absorption , which is an energy transfer to other than coherent re - radiation , displays a maximum
. a broadband lossy particle , with @xmath63 , which has only loss , does not absorb or scatter any light .
this is equivalent to a polarizability @xmath64 .
a perfect scatterer , with @xmath65 , does not absorb any light .
@xmath66 corresponds to the case @xmath65 and amounts to @xmath67 .
this is equivalent to the consideration of the cross sections in eqn .
[ formel14 ] .
due to energy conservation the whole extracted energy is re - radiated .
b ) streamlines of the time averaged energy flow of a perfect scatterer in a plane wave .
energy used for its excitation is re - radiated , such that on average no energy is absorbed i.e. in the time averaged picture we do not see streamlines ending up in the dipole .
c ) streamlines for a lossy system , the loss - parameter @xmath46 is set to 0.1 , the emitter is an effective sink of energy .
therefore , some streamlines end up in the emitter . here
, the @xmath34-plane is shown .
d ) same physical situation pictured in the @xmath33-plane .
the extensions of the cross - section @xmath68 and @xmath69 are not necessarily the same ( see also fig . [
fig:04]b ) . ] to visualize the energy flow we computed the time averaged poynting vector and solved the differential eqn .
[ formel22 ] numerically to get its streamlines .
@xmath70 with @xmath71 used as a `` dummy '' parameter .
such a picture was derived earlier , and allows a vivid insight on the process of light absorption @xcite . for the earlier discussed case , where no absorption occurs
, we show the case of streamlines in fig .
[ fig:03]b .
all streamlines are redirected , but do not end up in the dipole at the origin .
it must be noted that the streamlines leave the @xmath72-plane .
these points are marked with circles in the plot ( green ) .
the energy flow for the case of @xmath73 is shown in fig .
[ fig:03]c , d .
it is nicely visible how the dipole collects energy from a region , which is by far larger then the geometrical spread of the dipole ( a point - like dipole was considered ) .
moreover , it is visible that the streamlines are directed towards the dipole , even if they have already passed the position of the emitter .
the reason for that behavior is found in the interference terms .
the interfere happens in such a way that the energy flow is directed towards the dipole . for a lower @xmath46 value , the effect on
the incident plane wave is smaller , as evident on the @xmath72-coordinates . with a @xmath46 of unity
the distortion of streamlines exceeds one wavelength .
this effect is diminished with a lower @xmath46 .
+ based on the eqn .
[ formel11],b , c we see that @xmath65 refers to the ideal system were we have no absorption and the whole energy is re - radiated , whereas the @xmath63 case refers to no excitation , respectively no polarizability of the system .
[ fig:03]a shows the normalized energy contributions given in eqn .
[ formel11],b , c plotted as a function of @xmath46 .
the figure shows nicely the fact that absorption is necessarily concomitant with scattering i.e. absorption without scattering can not exist .
moreover , it occurs that the maximum value of absorption is given if @xmath62 , which is equivalent to the same amount of scattered and absorbed energy . as shown in ref .
@xcite , this is also valid for the case of @xmath74 , i.e. out of resonance . + .
it is visible that the streamlines leave the @xmath34-plane . by finding the boundary between the lines ending up in the dipole and the ones which are bypassing , we determine the effective absorption aperture .
b ) the shape of this absorption absorption aperture ( equivalent to the cross section ) for different values of @xmath46 far away from the emitter ( @xmath35=-15@xmath75 ) . the crossed labels on the curve @xmath76 in fig . [ fig:03]a correspond to the derived areas for the different @xmath46-values in here . ] based on the eqns .
[ formel11],b , c it is possible to define three different on resonant cross sections ( it is also possible to find a general description by using eqn .
[ formel7 ] ) @xmath77 as implied in fig .
[ fig:03]c , d the area of streamlines ending up in the singularity ( given by the dipole ) suggested a relationship to the absorption cross section .
for the three dimensional case , this is depicted in fig .
[ fig:04]a .
this definition represents a graphical association to eqn .
13c . based on this definition the shape of the absorption cross section was derived for different loss parameters .
the result is shown in fig .
[ fig:04]b . to derive the absorption cross section we iteratively searched for the point on the @xmath78-axis where a small deviation from the point @xmath79 yields a change of the streamlines whether they end up in the singularity or not .
once this point is found , the boundary of the aperture can be followed by making circles around the last derived point and checking the endpoint of the streamlines .
such apertures were calculated in the past @xcite .
the area of the apertures can than be derived by interpolating the data and using greens theorem . as visible from the plot ( fig .
[ fig:04]b ) the shape of the absorption cross section must not necessarily be round .
moreover , it has to be noted that while the theoretical provided behavior of the absorption cross section as a function of @xmath46 ( blue curve in fig .
[ fig:03]a ) would intuitively yield an assumption of a shape symmetry against a value of @xmath62 , the shape totally differs between a pair of @xmath46-values symmetric to 0.5 ( e.g. @xmath80 and @xmath81 in fig .
[ fig:04]b ) .
but the computation of the area of the absorption cross section based on the data of the simulation shows a good correspondence to the theoretical forecast ( black `` + '' in fig .
[ fig:03]a , i.e. the area of the pair @xmath80 and @xmath81 is equal ) .
the physical interpretation of this behavior implies that the amount of energy absorbed by a pair of loss parameters symmetric to 0.5 is equal , but the way the energy `` travels '' into the dipole is very different . for @xmath65 , i.e. no absorption , we find the textbook result for the extinction respectively scattering cross section of a dipole given by @xmath82 thus , the whole extinction i.e. the reduction of the incident power in the forward direction is due to scattering .
leaving the point of an ideal system ( without absorption ) the absorption cross section is of interest as well .
the maximum absorption cross section can be found for a loss factor of 0.5 given by @xmath83 again consistent with the result of @xcite .
so far , we did not introduce any polarization dependent response of the system , and one should note that the cross sections are indeed independent of the incident wave polarization ( i.e. linear or circular ) .
the expression for the maximum absorption cross section also holds for the off resonance case @xcite .
thus , a maximum energy dissipation is generally given if @xmath84 , which is also visible in fig .
[ fig:03]a .
the results which are presented for a hertzian dipole are equivalent to the analysis of e. shamonina and coworkers , who evaluated the energy flow into a short antenna in @xcite .
the equivalence of the results is not surprising since the inverse polarizability of a short antenna is given by @xcite @xmath85 thus , the math for a short antenna is included in eqn .
[ formel7 ] .
the parameter of interest in ref .
@xcite is @xmath86 which corresponds to @xmath46 via @xmath87 , i.e. the maximal absorption case in this nomenclature is given by @xmath88 .
the shape , we found for the absorption cross section of a hertzian dipole is equivalent to the results of ref . @xcite .
based on the analytically expression of the cross sections given by eqn .
13a , b , c one can easily disregard , that the definition of the cross section is just strictly valid for @xmath89 . a simple argument to understand this ,
is that the cross sections are defined via the incident power density @xmath90 .
since the total field is given by a superposition of the incident and the scattered field , the power density of the total field differs from the power density of the incident field .
thus , the cross sections given by eqn .
13a , b , c define an area which yields to the total amount of extinguished , scattered and absorbed power if they are multiplied with @xmath90 . hence , if the energy density of the total field changes the area of the cross section has to change to ensure that the amount of power which is extinguished , scattered or absorbed stays the same .
if the total energy density is e.g. smaller than the incident power density , the area of the cross sections has to increase , such that energy conservation is fulfilled .
[ fig:05 ] shows the qualitative behavior of the situation described above .
the absorption aperture of a dipole with a loss factor of 0.75 is plotted along the way to the dipole , furthermore some boundary solutions on and beside the aperture are shown .
at the situation `` far '' away from the dipole , where the power density is given by the incident power density , the shape of the aperture stays constant as expected ; approaching the emitter the shape changes . , of 0.75 is assumed , as in fig .
[ fig:04]a . ] based on the idea fig .
[ fig:05 ] provides , a more detailed investigation of the aperture in the optical near - field was performed .
the absorption aperture for different loss parameters was evaluated dependent on the distance to the dipole .
the results are shown in fig .
[ fig:06]a .
the black curve in the plot represents the theoretical curve given by eqn .
19c . as shown in the plot the area of the aperture increases in the optical near - field .
moreover , it occurs , that there is a maximum value of the aperture for a certain distance in front of the dipole . following these arguments , this
directly yields a change of the total power density . as an example
the situation for a loss parameter of @xmath91 is illustrated in fig .
[ fig:06]b .
the density of the energy flow is visualized as a grid of streamlines which were followed on their way towards the dipole .
the grid is color - coded corresponding to the energy density of the total field at the position .
for the situation in the @xmath72-plane at @xmath35=-6 @xmath92 $ ] the energy density appears uniform . for the corresponding situation in fig .
[ fig:06]a we realize , that the area nearly corresponds to the theoretical expectation in the far - field .
for the situation where a maximum in the aperture area occurs ( @xmath93 - 1.2 @xmath92 $ ] c.f . fig .
[ fig:06]a ) , the power density is reduced .
thus , the simulation is conformant to the previous argument . .
the dashed parts of the curves could not be calculated .
b ) energy density inside the absorption cross section .
we assume a rectangular grid , which is followed along the @xmath35-axis towards the emitter . it is evident , that the cross section initially widens , and it is then reshaped . as an example the energy density at @xmath94 is obviously reduced , whereas the overall area is enlarged .
the corresponding situation in a ) is marked with a gray dashed line . ]
[ fig:06]b shows , that the power density does not change homogeneously approaching the emitter .
close to the emitter , the shape of the grid adapts the shape of the dipole radiation pattern ( donut - shape ) .
this indicates , that the energy flow in the near - field is dominated by the energy flow of the dipole itself .
this corresponds to the result of g. zumofen and coworkers @xcite , when they describe that the dipolar component of an incident focused wave can be perfectly reflected by a single dipole . in our description ,
the energy flow close to the dipole seems to match such a dipolar pattern .
a ) shows the energy flow for @xmath95 for selected points inside and outside the aperture .
the same points are followed in b ) for the case of @xmath96 .
the size of the cross section is the same .
c ) and d ) show the energy flow for points slightly inside and outside of the aperture .
although the aperture is the same , the energy flow differs in the optical near - field .
these pictures correspond to the calculations in @xcite , but show the energy flow _ into _ the dipole . ]
if one considers an atom as a single emitter it is known from atomic physics that linear as well as circular polarized excitation is of interest .
based on the math introduced in section 2 it is simple to formulate the problem for a circular excitation . with an incident wave given by @xmath97 where @xmath98 corresponds to the jones vector of a circular polarized wave .
this yields to a dipole moment ( on resonance ) @xmath99 notice the common phase shift of @xmath37 between the incident wave and the reaction of the driven oscillator .
the eqns .
[ formel23a ] and [ formel24 ] correspond to the eqns .
[ formel2 ] and [ formel9 ] in section 2 . based on this modification the same investigations as for the hertzian dipole
were performed .
[ fig:07 ] shows the apertures as well as some streamlines for the circular case and different loss parameters . the way the energy travels into the dipole corresponds to the energy flow of a circular dipole emitter itself ( without the incident beam ) as investigated in @xcite .
one has to notice , that the direction of the energy flow in our case is towards the dipole while the energy flow of a radiating dipole is logically outwards .
this is due to the interference terms which direct the energy into the dipole . if the plots in fig .
[ fig:07]a , b are compared , it is nicely visible that the behavior of the energy flow is more dominated by the scattered part for @xmath61 .
if one follows the solution which bypasses the dipole it occurs that the solution first spins around the @xmath35-axis until it changes the direction and gets sucked up into the dipole .
[ fig:07]c , d shows the boundary solutions for the two cases plotted in fig .
[ fig:07]a , b . the pair of loss values ( @xmath46 ) chosen for the plots in fig .
[ fig:07 ] are symmetric to 0.5 considering the theoretical basis ( c.f .
[ fig:03]a/ fig .
[ fig:08]a ) that means the amount of energy absorbed by the dipole is equivalent for both cases . as shown in fig .
[ fig:08]a the simulation data satisfies the theoretical forecast ( which is exactly the same as for the linear case ) for a circular excitation as well .
but if the results are compared to the results of a hertzian dipole it occurs that there is a shape symmetry against a loss value of 0.5 ( fig .
[ fig:08]b ) .
this is more intuitive than for the hertzian dipole case if one has the symmetric theory plot in mind ( c.f .
[ fig:03]a ) .
moreover , it occurs that the shape of the absorption cross section stays perfectly round as visible in fig .
[ fig:08]b . investigating the energy flow of a symmetric pair of loss values
it appears that the way the energy takes differs drastically for both cases ( see also fig .
[ fig:07 ] ) .
the cases corresponding to the plots in b ) are highlighted with a bold + .
b ) apertures , far away from the dipolar emitter for different values of the loss parameter @xmath46 .
the determined values in a ) correspond to calculations for different apertures . unlike for the hertzian dipole
, we observe a shape symmetry against the value of @xmath62 , although the energy flow is different , as shown in fig .
[ fig:07 ] . ] comparing the situations for the hertzian and the circular dipole it is visible that both mathematical descriptions nicely yield a result which satisfies the underlying theory , independent of the incident wave polarization . whereas the energy flow for the different cases differs totally , but it always fulfills the boundary conditions that the `` total area '' of the absorption cross section yields to a corresponding result .
thus , it might be questionable how meaningful the energy flow itself is . especially , since the definition of the poynting vector is not unique .
but the results of the simulation provided in the current paper may give an intuitive picture how light absorption occurs .
for the hertzian dipole , it was determined , that the effective far - field aperture is altered in the optical near - field .
this seems to be equivalently the case for the circular dipole .
an example is depicted in fig .
[ fig:08]a .
these apertures are relatively easy to determine , since their shape is always circular , such that the derivation of one point on the boundary of the aperture is sufficient to calculate the area ( @xmath100 ) .
the presented results describe the situation for a plane wave and a dipole emitter , and are fully equivalent for short antennas in the rf- or mw - range .
the emitter can be a circular or a linear point - size dipole . to determine the absorption cross - section ,
the streamlines of the poynting vector are followed , and result in an effective aperture , which matches the analytical derivations . for a point - like hertzian dipole ,
the apertures are not circularly symmetric .
the absorption cross - section is fully equivalent in both cases .
whereas for the linear case , the effective aperture changes its shape with the loss parameter , @xmath46 , the circular dipole always has a circular shape . in both cases ,
the apertures are changed in the optical near - field .
this might not necessarily be interesting for experiments with focused light .
combinations with nano - particles or plasmonic structures might explore the described effects .
extinction experiments in the optical near - field @xcite will be likely dominated by other effects , such as the emission of sub - wavelength apertures .
the described math is presently extended to the case of multiple dipolar emitters , which might `` cloak '' each other by their presence .
this yields situations comparable to ref . @xcite .
furthermore , we extend our approach to the case of a real optical focus , as described earlier @xcite .
this would correspond to a graphical representation of the case described by ref .
i.g . acknowledges the discussions with g. wrigge and g. zumofen ( eth zrich ) regarding this topic .
we acknowledge funding from the max planck - ubc quantum materials center , the max planck society ( j.w . , max planck fellowship ) , and the eu via the project siqs .
the source code ( ` mathematica 9.0.1 , wolfram research ` ) to derive fig .
[ fig:03]b , c , d and the effective aperture shape can be found on http://gerhardt.ch/apertures.php[`http://gerhardt.ch/apertures.php ` ] .
k. g. lee , x. chen , h. eghlid , p. kukura , r. lettow , a. renn , v. sandoghdar , and s. gtzinger . a planar dielectric antenna for directional single - photon emission and near - unity collection efficiency .
, 5(3):166169 , march 2011 .
vamivakas , m. atatre , j. dreiser , s.t .
yilmaz , a. badolato , a.k .
swan , b.b .
goldberg , a. imamoglu , and m.s . nl .
strong extinction of a far - field laser beam by a single quantum dot .
, 7(9):28922896 , 2007 .
meng khoon tey , zilong chen , syed abdullah aljunid , brenda chng , florian huber , gleb maslennikov , and christian kurtsiefer .
strong interaction between light and a single trapped atom without the need for a cavity . , 4(12):924927 , december 2008 .
meng khoon tey , gleb maslennikov , timothy c h liew , syed abdullah aljunid , florian huber , brenda chng , zilong chen , valerio scarani , and christian kurtsiefer . interfacing light and single atoms with a lens .
, 11:043011 , 2009 . | the optical interaction of light and matter is modeled as an oscillating dipole in a plane wave .
we analyze absorption , scattering and extinction for this system by the energy flow , which is depicted by streamlines of the poynting vector .
depending on the dissipative damping of the oscillator , the streamlines end up in the dipole . based on a graphical investigation of the streamlines ,
this represents the absorption cross section , and forms a far - field absorption aperture . in the near - field of the oscillator , a modification of the aperture is observed .
this scheme can be adapted to a single dipolar emitter , interacting with a light field . in the case of the absorption by a single atom , where the oscillator has a circular dipole characteristics , we model the energy flow and derive the apertures . |
a _ binary relation _ on a set @xmath1 is a subset @xmath4 of the cartesian product @xmath5 .
we write @xmath6 instead of @xmath7 .
a map @xmath8 _ preserves _
@xmath4 if @xmath9 for all @xmath10 .
a _ binary system _ on the set @xmath1 is a pair @xmath11 where each @xmath12 is a binary relation on @xmath1 .
an _ endomorphism _ of @xmath0 is any map @xmath8 preserving each @xmath12 .
the identity map on @xmath1 is an endomorphism of @xmath0 .
if there is no other endomorphism , @xmath0 is _
rigid_. provided that the relations @xmath12 are reflexive , the constant maps are endomorphisms , too .
we say @xmath0 is _ semirigid _ if the identity map and the constant maps are the only endomorphisms .
rigidity and semirigidity have attracted some attention ( eg see @xcite , @xcite , @xcite ) .
systems of equivalence relations lead to prototypes of semirigid systems .
indeed , as mentioned by r. s. pierce @xcite ( problem 2 , p. 38 ) , _ if a set @xmath1 has at least three elements , only the constant functions and the identity map preserve all equivalence relations on @xmath1_. from this , it follows that : [ pierce ] if a set @xmath13 of equivalence relations generates by means of joins and meets ( possibly infinite ) the lattice @xmath14 of equivalences relations on @xmath1 , then @xmath15 is semirigid
. the converse does not hold . indeed , according to strietz @xcite ,
if @xmath1 is finite with at least four elements , four equivalences are needed to generate @xmath14 by joins and meets whereas , as shown by zdori ( 1983 ) @xcite , for every set @xmath1 , whose size @xmath16 is finite and distinct from @xmath2 and @xmath3 , there is a semirigid system made of three equivalence relations .
result reads as follows .
[ zadori ] let @xmath17 .
the following system of three equivalence relations @xmath18 on @xmath19 is semirigid . 1
. case @xmath20 , @xmath21 : @xmath22 2 .
case @xmath23 , @xmath24 : @xmath25 the case @xmath26 even is represented by the colored graph of figure [ zadori ] . in this graph ,
connected components formed by single colored edges represent the blocks of a partition into equivalence classes . for @xmath26 odd , simply delete the node 0 which is located on top of the graph and relabel conveniently the vertices . in @xcite we investigated semirigid systems .
in this paper we describe a general construction of semirigid systems of three equivalence relations which includes zdori s .
let @xmath27 denote the set of real numbers and @xmath28 its cartesian square .
we define three equivalence relations @xmath29 and @xmath30 on @xmath28 .
we denote by @xmath31 and @xmath32 the first and second projections from @xmath33 onto @xmath27 and by @xmath34 the map @xmath35 .
next , for @xmath36 , we define @xmath37 as the kernel of @xmath38 , _
@xmath39 for all @xmath40 such that @xmath41 . finally , we set @xmath42 .
the system @xmath43 is , by far , not semirigid .
but , there are many subsets @xmath44 of the plane for which the system @xmath45 induced by @xmath43 on @xmath44 is semirigid . in section [ section : abelian ]
we introduce the notion of _ monogenic subset _
( see definitions [ def : monogenic ] given in subsection [ subsection : monogenic ] ) and with the simple notion of symmetry in the plane we prove : [ maintheo ] if a finite subset @xmath44 of @xmath28 is monogenic and has no center of symmetry in @xmath28 then @xmath45 is semirigid .
a simple example , which is at the origin of this result , is the following .
let @xmath46 denotes the set of non - negative integers
. for @xmath47 set @xmath48 .
then @xmath49 satisfies the hypotheses of theorem [ maintheo ] .
hence @xmath50 is semirigid .
this yields an example of a semirigid system of three pairwise isomorphic equivalence relations on a set having @xmath51 elements .
now , set @xmath52 and @xmath53 .
both sets satisfy the hypotheses of theorem [ maintheo ] , hence the induced systems @xmath54 and @xmath55 are semirigid .
as it is easy to see , these two examples are isomorphic to those of zdori . with the three equivalence relations , width=480 ] in the vein of theorem [ maintheo ] we prove the following result ( see proposition [ prop .
semirigid ] for a more precise statement ) .
[ thm : infinite]for each cardinal @xmath56 or @xmath57 , there exists a semirigid system of three equivalences on a set of cardinality @xmath58 .
our study leads to several questions that we mention in the paper . among these
are the following .
[ problems ] 1 .
does the conclusion of theorem [ thm : infinite ] extend to every cardinal @xmath59 ?
2 . a complete description or a useful characterization of semirigid systems is not known .
find algorithms deciding in reasonable time whether a system of @xmath58 equivalence relations on a set of size @xmath26 is semirigid or not .
the algorithmic complexity of the problem does not seem to be known .
systems of equivalence relations leading to theorem [ maintheo ] are subsystems of systems of equivalence relations associated with three directions in the plane .
describe finite semirigid systems which are embeddable into the system associated with three directions in the plane .
describe a test to decide whether a systems of three equivalence relations is embeddable into a system of equivalences associated to three directions in the plane .
concerning item @xmath60 and @xmath61 of problems [ problems ] let us mention that the class of systems embeddable into a system of equivalences given by three directions in the plane is closed under embeddability .
this fact leads to he following problems which are the finite minimal non representable systems ?
how many non isomorphic are they ?
finitely many ? to conclude , we indicate first how our study fits in the context of universal algebra .
the set of congruences of a universal algebra on a set @xmath1 is a basic example of a set of equivalence relations .
this set ordered by inclusion forms a lattice , in fact a sublattice of the lattice @xmath14 .
one of the oldest unsolved problem in universal algebra is the finite lattice representation problem : _ is every finite lattice isomorphic to the congruence lattice of a finite algebra ? _ ( see @xcite ) .
an approach to this problem is related to our study . indeed , for @xmath62 a subset of @xmath14 , let @xmath63 , resp .
@xmath64 , be the set finitary operations , resp .
unary operations , on @xmath1 which preserve all members of @xmath62 and for a set @xmath65 of finitary operations on @xmath1 , let @xmath66 be the set of equivalence relations preserved by all members of @xmath65 . ordered by inclusion , the set @xmath66 is a lattice , the congruence lattice @xmath67 of the algebra @xmath68 . according to a result of a.maltsev , @xmath69 is determined by its unary part @xmath70 , hence the above problem amounts to prove that every finite lattice @xmath71 is isomorphic to the congruence lattice of an algebra defined by unary operations . in general
, a sublattice @xmath71 of @xmath14 is not the congruence lattice of an algebra on @xmath1 . indeed ,
pierce s result implies that relational system @xmath72 made of equivalence relations is semirigid if and only if @xmath73 ( where @xmath74 and @xmath75 ) .
we recall that @xmath76 is the lattice on five element made of a @xmath77-element antichain and a top and bottom added .
three equivalences together with the equality relation and the full equivalence relation on @xmath1 form an sublattice of @xmath14 isomorphic to @xmath78 if @xmath79 the meet of any two distinct equivalence relations is the equality relation ( this amounts to the fact that the intersection of any pair of equivalence classes belonging to two distinct equivalence relations contains at most one element ) and @xmath80 the join of two is the full set @xmath81 of ordered pairs .
as shown by zdori @xcite : [ thm : zadori2 ] if @xmath82 made of three equivalences relations is semirigid then these equivalence relations together with the equality relation and the full equivalence relation on @xmath1 form a sublattice @xmath71 of @xmath14 , isomorphic to the lattice @xmath78 . provided that @xmath1 has more than three elements , the lattice @xmath71 , mentionned in theorem [ thm : zadori2 ] , is distinct from @xmath83 ( which is equal to @xmath14 since @xmath43 is semirigid ) hence is not the congruence lattice of an algebra on @xmath1 .
so our study of semirigidity is somewhat opposite to the representability question mentioned above .
still , the semirigidity notion appears ( briefly ) in the thesis of w. demeo @xcite devoted to the representation of finite lattices .
the sublattices @xmath71 of @xmath14 such that @xmath84 ( where @xmath85 ) are said to be _ dense _ ; the fact that , as a lattice , @xmath78 has a dense representation in every @xmath14 with @xmath1 finite on at least five elements , amounting to zdori s result , appears in @xcite as proposition 3.3.1 on page 20 . according to @xcite ,
if @xmath62 is a set of equivalences on a finite set @xmath1 , the congruence lattice @xmath83 is made of all equivalence relations definable by primitive positive formulas from @xmath62 ( see @xcite for an easy to read presentation ) .
we hope that this result could help in our problems on semirigid relations .
next , we mention that some systems made of three equivalence relations on the same set play an important role in the study of latin squares and quasigroups .
indeed , behind these objects are @xmath77-nets , objects usually presented in terms of incidence structures rather than equivalence relations .
a _ @xmath77-net of order @xmath26 _ is a point - line incidence structure made of @xmath86 points and three classes of @xmath26 lines such that : 1 .
any two lines from different classes are adjacent ; 2 .
no two lines from the same class are adjacent ; 3 .
any point is incident with exactly one line from each class .
as the reader will easily observe , this amounts to give three equivalence relations @xmath12 , @xmath87 , on the set @xmath1 of points such that each equivalence class of @xmath12 intersects in exactly one element each equivalence class of any other @xmath88 .
let @xmath12 , @xmath87 , be three equivalence relations on a set @xmath1 which satisfy this condition ( we do not demand on@xmath1 to be finite ) ; let @xmath89 be the set of equivalence classes of @xmath12 and @xmath90 be the canonical map .
let @xmath91 be defined by @xmath92 where @xmath93 is the unique element of the intersection @xmath94 .
then the partial maps @xmath95 and @xmath96 are bijective for every @xmath97 and @xmath98 .
conversely , any map @xmath99 satisfying these conditions comes from a @xmath77-net . if the elements of the sets @xmath89 are identified to those of a set @xmath100 , by means of three bijective maps @xmath101 we may then define a quasigroup operation @xmath102 on @xmath100 ( see subsection [ quasigroup ] ) by setting @xmath103 .
if @xmath100 is an @xmath26-element set , e.g. @xmath104 , the @xmath26 by @xmath26 matrix whose coefficient @xmath105 is equal to @xmath106 is the _ table of the quasigroup_. any table of this form is a _ latin square on the symbols @xmath107_. a given @xmath77-net yields several quasigroups which are said _
isotopic_. hundreds of papers and several books , notably @xcite , have been published on latin squares , quasigroups and @xmath77-nets .
we mention few facts connected to the problems we are considering .
the _ graph of a @xmath77-net _ @xmath108 is the graph @xmath109 with vertex set @xmath1 and edges the pairs of distinct vertices which belong to some equivalence class of some @xmath12 .
the _ graph of a latin square _ is the graph of the corresponding @xmath77-net .
phelps @xcite has shown that for each integer @xmath110 there is a latin square of order @xmath26 whose graph has no proper automorphism .
no ( reflexive ) graph with more than a vertex can be semirigid , but it remains to be seen whether the @xmath77-nets of phelps are semirigid .
does there exist a semirigid @xmath77-net of order @xmath111 for every cardinal @xmath58 ( finite or not ) larger or equal to @xmath112 ?
if @xmath0 is a @xmath77-net , the equivalence relations @xmath113 together with the equality relation and the full equivalence relation on @xmath1 , form a sublattice of @xmath14 , isomorphic to the lattice @xmath76 hence satisfy conditions @xmath79 and @xmath80 beforetheorem [ thm : zadori2 ] above .
if @xmath0 is induced by a @xmath77-net on some subset then it satisfies condition @xmath79 .
conversely a system @xmath0 satisfying condition @xmath79 can be extended to a system @xmath114 which forms a @xmath77-net .
furthermore , if @xmath1 is finite one can find @xmath114 with order at most @xmath115 ( proposition [ prop : extending ] ) . in the finite case
, this is a rewording of a result about extension of partial latin squares due to evans @xcite . in the infinite case
, this follows from compactness theorem of first order logic . despite some possible confusion ,
we call _ partial @xmath77-net _ any system induced by some @xmath77-net .
the addition on the set @xmath27 of real numbers is a prototypal example of a quasigroup operation .
the @xmath77-net associated with this operation is made of the three equivalence relations @xmath29 and @xmath30 on @xmath28 .
our results are about partial @xmath77-nets induced by this @xmath77-net .
this paper is divided into two more sections . in the first section , we recall the definitions that we need .
we indicate that systems of equivalence relations are pre - ultrametric spaces , a fact which may also motivate the study of the semirigidity of these systems .
we give a representation theorem for reduced systems of equivalence relations ( theorem [ theo1 ] ) .
we end the section with some examples of semirigid systems ; theorem [ theo3 ] contains , with an other proof , the fact that @xmath50 is semirigid .
results announced in the introduction are proved in the last section that is self - contained up to subsection [ subsection : self - contained ] .
an _ equivalence relation _ on a set @xmath1 is a binary reflexive , symmetric and transitive relation on @xmath1 .
system of equivalence relations _ on @xmath1 is a pair @xmath11 where each @xmath12 is an equivalence relation on @xmath1 .
if @xmath65 is a subset of @xmath1 , the _ restriction _ of @xmath0 to @xmath65 is @xmath116 . if @xmath117 is another system of equivalence relations on @xmath118 , a _ homomorphism _ from @xmath0 to @xmath114 is a map @xmath119 such that for all @xmath120 : @xmath121 if the map @xmath122 is injective and he implication in ( [ eq:1 ] ) is an equivalence
then @xmath122 is an _ embedding _ ; if furthermore @xmath122 is bijective , @xmath122 is an _ isomorphism _ and the systems @xmath0 and @xmath114 are _
isomorphic_. a homomorphism of @xmath0 into @xmath0
is called an _ endomorphism _ of @xmath0 ; the set of endomorphisms of @xmath0 is denoted by @xmath123 .
a system of equivalence relations @xmath124 is _ reduced _ if @xmath125 , where @xmath126 .
note that _
if @xmath0 is semirigid then it is reduced_. indeed , supposing that @xmath127 is distinct from @xmath128 , one choose some @xmath129 .
then the map @xmath122 defined by @xmath130 and @xmath131 for @xmath132 is an endomorphism of @xmath0 . hence @xmath0 is not semirigid .
propositions [ prop : ultra1 ] and [ prop : ultra2 ] of this subsection express the fact that systems of equivalence relations and pre - ultrametric spaces are two faces of the same coin .
the study of metric spaces with distance values in a boolean algebra first appears in blumenthal @xcite .
a general study of distances with values in an ordered set is in @xcite .
ultrametric spaces with distance values in an ordered set have been studied by priess - crampe and ribenboim in several papers , e.g. @xcite , @xcite,@xcite .
a _ join - semilattice _ is an ordered set in which two arbitrary elements @xmath97 and @xmath98 have a join , denoted by @xmath133 , defined as the least element of the set of common upper bounds of @xmath97 and @xmath98 .
let @xmath100 be a join - semilattice with a least element , denoted by @xmath134 .
a _ pre - ultrametric space _ over @xmath100 is a pair @xmath135 where @xmath136 is a map from @xmath5 into @xmath100 such that for all @xmath137 : @xmath138 the map @xmath136 is an _ ultrametric distance _ over @xmath100 and @xmath139 is an _ ultrametric space _ over @xmath100 if @xmath139 is a pre - ultrametric space and @xmath136 satisfies _ the separation axiom _ : @xmath140 given a set @xmath141 let @xmath142 be the power set of @xmath141 .
then @xmath142 , ordered by inclusion , is a join - semilattice ( in fact a complete boolean algebra ) in which the join is the union , and @xmath134 the empty set .
[ prop : ultra1 ] let @xmath11 be a system of equivalence relations . for @xmath10 , set @xmath143 .
then the pair @xmath144 is a pre - ultrametric space over @xmath142 .
conversely , let @xmath135 a pre - ultrametric space over @xmath142 . for every @xmath145 set @xmath146 and let @xmath11 .
then @xmath12 is an equivalence relation on @xmath1 and @xmath147 .
furthermore , @xmath148 is an ultrametric space if and only if @xmath0 is reduced . for a join - semilattice @xmath100 with a @xmath134 and for two pre - ultrametric spaces @xmath135 and @xmath149 over @xmath100 , a _ non - expansive mapping _ from @xmath139 to @xmath150 is any map @xmath151 such that for all @xmath152 : @xmath153 [ prop : ultra2 ] let @xmath11 and @xmath154 be two systems of equivalence relations .
a map @xmath155 is a homomorphism from @xmath0 into @xmath114 if and only if @xmath122 is a non - expansive mapping from @xmath148 into @xmath156 .
propositions [ prop : ultra1 ] and [ prop : ultra2 ] are immediate and the proofs are left to the reader .
still , this suggests to study ultrametric spaces ( in the ordinary sense ) which are rigid with respect to non - expansive mappings .
but notice that a non - trivial ultrametric space could not be semirigid .
indeed : an ultrametric space @xmath157 over a linearly ordered set @xmath100 having at least two elements is not semirigid .
if all non - zero distances between elements of @xmath139 are equal then every permutation of @xmath100 is an endomorphism of @xmath158 , hence @xmath139 is non semirigid .
thus assume that there are @xmath159 in @xmath100 and @xmath160 such that @xmath161 .
let @xmath162 .
then @xmath163 .
indeed , if @xmath164 then @xmath165 . hence ,
@xmath166 . let @xmath122 be the map defined by setting @xmath167 if @xmath168 and @xmath169 otherwise .
then @xmath122 is non - expansive .
indeed , let @xmath170 .
suppose @xmath171 and @xmath172 . since @xmath100 is totally ordered , @xmath173 . since @xmath174 , we have @xmath175 .
since @xmath100 is totally ordered , we also have @xmath176 .
this amounts to @xmath177 .
the other cases lead trivially to the same inequality .
hence , @xmath122 is non - expansive . beyond the fact that it is not linearly ordered , what can be said about the order structure of the set of values of the distances of a non - trivial semirigid ultrametric space ?
contrary to the case of ultrametric spaces , there exist metric spaces ( in the ordinary sense ) which are rigid with respect to non - expansive maps .
in fact there are metrizable topological spaces which are rigid with respect to continuous maps .
indeed , as shown by de groot @xcite , there exist connected and locally connected subsets of the plane which are semirigid with respect to continuous maps ( see corollary 3 in @xcite ) .
describe the semirigid metric spaces with respect to non - expansive mappings .
we may also represent systems of equivalence relations on finite sets by matrices .
let @xmath178 be an @xmath179 by @xmath26 matrix with non - negative integer coefficients .
for each @xmath180 let @xmath88 be the equivalence relation defined on @xmath181 by @xmath182 then @xmath183 is a system of equivalence relations .
conversely : [ lem ] every system of @xmath26 equivalence relations on an @xmath0-element set is isomorphic to a system @xmath184 for some @xmath179 by @xmath26 matrix @xmath19 with non - negative integer coefficients .
let @xmath0 be a system of @xmath26 equivalence relations on an @xmath179-element set .
without loss of generality we may suppose that the @xmath179-element set is @xmath181 and hence that @xmath185 .
for @xmath186 and @xmath187 set @xmath188 .
let @xmath189 be a one - to - one map from the set @xmath190 into the set of non - negative integers .
set @xmath191 . what is the complexity of the problem :
decide whether or not the system of equivalence relations associated with an @xmath179 by @xmath26 matrix is semirigid ? for reduced systems , lemma [ lem ] has a somewhat simpler form .
let @xmath15 be a sytem of equivalence relations .
for @xmath192 and for @xmath145 let @xmath193 be the equivalence class of @xmath97 and let @xmath194 be the set of equivalence classes .
let @xmath195 be a family of sets and let @xmath196 be their cartesian product . for @xmath197 denote by @xmath198 its @xmath199-th projection , hence @xmath200
let @xmath201 where @xmath202 if @xmath203 . clearly this system is reduced .
if all @xmath89 are equal to the same set @xmath204 , we denote the system by @xmath205 .
[ theo1 ] every reduced system can be embedded into a system of the form @xmath206 .
let @xmath15 .
let @xmath195 where @xmath194 is the set of equivalence classes of @xmath12 and let @xmath207 be defined by setting @xmath208 .
then , the map @xmath4 is an embedding from @xmath0 into @xmath209 : first , @xmath4 is one - to - one . indeed , suppose that @xmath210 for some @xmath10
. then @xmath211 , that is @xmath212 , for all @xmath145 . since @xmath0 is reduced , @xmath213 .
next , @xmath122 is an homomorphism of @xmath0 into @xmath209 .
suppose that @xmath214 for some @xmath215 and @xmath145 .
then @xmath212 , that is @xmath211 , which amounts to the fact that @xmath216 .
the proof of the converse is similar .
[ cor1 ] every reduced system can be embedded into a system of the form @xmath205 , where @xmath217 .
@xmath209 is embeddable into @xmath205 , where @xmath218 . indeed , to @xmath219 associate @xmath220 defined by @xmath221 .
let @xmath11 be a system of equivalence relations and @xmath122 be a selfmap of @xmath1 .
it is immediate to see that @xmath122 is an endomorphism of @xmath0 if and only if for each @xmath145 , the map @xmath122 induces a selfmap @xmath222 on the set @xmath89 of equivalence classes of @xmath12 .
furthermore , if @xmath0 is reduced , the family @xmath223 determines @xmath122 .
theorem [ theo2 ] below expresses this fact and characterizes families @xmath224 that come from an endomorphism .
let @xmath225 and for @xmath145 denote @xmath89 the image of the @xmath199-th projection of @xmath1 in @xmath204 , that is , with our notations , @xmath226 .
note that @xmath227 .
for @xmath228 , denote by @xmath229 the map from @xmath230 into @xmath231 defined by setting : @xmath232 [ theo2 ] let @xmath225 .
a map @xmath233 is a homomorphism of @xmath234 into @xmath205 if and only if @xmath235 for some @xmath228 .
moreover , if such an element @xmath236 exists it is unique .
suppose that @xmath122 is a homomorphism .
let @xmath145 and let @xmath237 .
let @xmath238 such that @xmath239 .
then @xmath240 . since @xmath122 is a homomorphism , @xmath241 , amounting to @xmath242 .
. clearly @xmath244 .
set @xmath245 .
we have @xmath235 .
the converse is immediate .
[ cor3]let @xmath225 .
then @xmath234 is semirigid if and only if for every @xmath246 such that for every @xmath192 , @xmath247 then either each @xmath248 is a constant function or each @xmath248 is the identity function on @xmath89 .
there is a bijective correspondence between @xmath249 and @xmath250 .
let @xmath251 .
then @xmath229 is an endomorphism of @xmath252 .
theorem [ theo2 ] ensures that every endomorphism has this form .
let @xmath253 .
then @xmath252 has @xmath254 endomorphisms .
each endomorphism @xmath122 of @xmath252 can be expressed as : @xmath255 where @xmath256 , @xmath257 , and the sum is modulo 2 .
[ theo3]let @xmath258 , @xmath47 and let @xmath141 be a set ( not necessarily finite ) with at least three elements .
set : @xmath259 then @xmath260 is semirigid .
observe first that @xmath261 for each @xmath145 .
apply corollary [ cor3 ] .
let @xmath246 be such that condition ( [ eq:3 ] ) of corollary [ cor3 ] holds .
[ claim1]if @xmath262 with @xmath263 then for all @xmath264 : @xmath265 indeed , let @xmath266 be distinct from @xmath199 and @xmath267 .
let @xmath268 and @xmath269 be defined by setting @xmath270 if @xmath271 , @xmath272 , @xmath273 and @xmath274 . then @xmath275 .
since condition ( [ eq:3 ] ) holds , @xmath276 and @xmath277 belong to @xmath1 , that is @xmath278 .
thus @xmath279 , proving our claim .
[ claim2 ] @xmath280 for all @xmath281 .
indeed , this equality holds for @xmath282 ; for larger @xmath93 apply induction and claim [ claim1 ] .
[ claim3 ] the value @xmath283 is independent of @xmath199 .
indeed , from claim [ claim1 ] we obtain @xmath284 .
if @xmath285 then according to claim [ claim2 ] , we have that @xmath286 for all @xmath199 and hence all maps @xmath248 are constant , proving that @xmath236 is constant .
thus , we may suppose @xmath287 .
[ claim4 ] @xmath288 and @xmath289 for all @xmath145 . indeed , from claim [ claim2 ]
, we have that @xmath290 . since @xmath291
, it follows that @xmath292 .
suppose that @xmath293
. then @xmath294 and @xmath295 for each @xmath145 .
since @xmath296 , the image @xmath297 of the sequence @xmath97 such that @xmath298 and @xmath299 for @xmath300 satisfies @xmath301 and @xmath302 . since @xmath296 ,
this image is not in @xmath1 .
this case is then impossible .
the only remaining possibility is @xmath303 . in this case ,
@xmath304 for all @xmath199 and @xmath93 .
according to corollary [ cor3 ] , @xmath260 is semirigid .
\1 ) set @xmath305 .
then , the map @xmath122 from @xmath1 into @xmath306 defined by setting @xmath307 for all @xmath308 a is an isomorphism from @xmath260 onto @xmath50 .
hence , @xmath50 is semirigid .
\2 ) let @xmath309 and @xmath296 . in this case
, @xmath1 is the set of characteristic functions of singletons , that is functions @xmath310 taking value @xmath311 on @xmath199 and @xmath134 on elements @xmath312 distinct from @xmath199 . for each @xmath199 ,
clearly @xmath313 is the equivalence relation which puts @xmath310 in one class and all other characteristic functions in an other class .
hence , this system of equivalence relations is isomorphic to @xmath314 , where @xmath315 if either @xmath316 or @xmath317 . the fact that the system @xmath0 is semirigid follows directly from lemma [ pierce ] .
indeed , for all @xmath318 , clearly @xmath319 is a co - atom in the lattice of equivalence relations , and all co - atoms are obtained this way .
thus the @xmath12 s generate the lattice of equivalence relations on @xmath141 .
the simplest systems of three equivalence relations are those associated to groups .
their endomorphisms are described in theorem [ cor : additivemaps ] .
these systems are not semirigid , but some induced systems are . in order to obtain some examples of semirigid systems we describe in lemma [ lem : additive ] the endomorphisms defined on induced systems .
let @xmath320 be a group , the operation being denoted by @xmath321 and the neutral element @xmath311 . recall that a _ group homomorphism _ is any map @xmath322 from @xmath320 into @xmath320 such that @xmath323 for all @xmath324 .
set @xmath325 .
denote by @xmath31 and @xmath32 the first and second projections from @xmath1 onto @xmath320 and let @xmath326 be defined by @xmath327 ( hence @xmath328 ) . for @xmath87 , let @xmath37 denote the equivalence relation on @xmath1 defined for all @xmath329 by setting @xmath39 if @xmath41 . finally , set @xmath330 and let @xmath44 be a subset of @xmath1 .
[ lem : additive ] a map @xmath331 is a homomorphism from @xmath332 into @xmath0 if and only if for each @xmath87 there is a map @xmath333 such that @xmath334 for all @xmath335 .
in particular , these three maps satisfy : @xmath336 for all @xmath335 . the first part of the lemma holds if @xmath320 is a _ magma _
( rather than a group ) , that is just a set equipped with a binary operation ; the second part holds if in addition this operation has a neutral element , that is an element , denoted by @xmath311 , and necessarily unique , such that @xmath344 for all @xmath345 . since each @xmath346 identifies with the canonical projection onto the quotient of @xmath347 , a homomorphism @xmath122 from @xmath332 into @xmath0 induces three maps @xmath348 , @xmath349 ( cf .
the proof of theorem [ theo2 ] ) .
the fact that equation ( [ eq : additivity ] ) holds follows readily from the definitions . indeed , for @xmath335
, we have : @xmath350 @xmath351 for the second part , let @xmath352 .
the relation @xmath353 yields @xmath354 hence @xmath355 ; similarly , @xmath356 .
the relation @xmath357 yields @xmath358 thus @xmath359 . from @xmath360
we obtain @xmath361 .
applying equation ( [ eq : additivity ] ) to @xmath97 and @xmath311 gives @xmath362 .
hence @xmath363 , as claimed .
due to the associativity of @xmath321 , if @xmath122 has the form given above , then it is a homomorphism of @xmath0 . in particular , for every @xmath367 , the transformation @xmath368 defined on @xmath1 by setting @xmath369 is an endomorphism of @xmath0 .
now , let @xmath122 be an endomorphism of @xmath0 .
let @xmath370 and @xmath371 .
since these transformations are endomorphisms of @xmath0 , the map @xmath372 is an endomorphism of @xmath0 fixing @xmath337 . according to lemma [ lem : additive ] ,
@xmath373 is of the form @xmath374 where @xmath322 is a homomorphism of @xmath320 .
since @xmath375 , @xmath122 has the form given above . a _
quasigroup _ is a magma @xmath376 of which each translation is bijective ( each element @xmath377 yields the left translation @xmath378 and the right translation @xmath379 ) .
let @xmath320 be a quasigroup .
on @xmath380 , the system @xmath330 defined as in the case of groups is a @xmath77-net .
the first part of lemma [ lem : additive ] applies and , if @xmath320 has an identity , the second part too .
a quasigroup @xmath376 with an identity is a _
loop_. it is known that a semigroup @xmath320 satisfying the identity @xmath381 for all @xmath382 is a loop , called a _
moufang loop_. given an @xmath384 , let @xmath385 be defined by setting @xmath386 .
this map is an endomorphism of @xmath0 .
it is not constant since @xmath320 is a quasigroup .
it is not the identity since @xmath387 .
since it is not the identity , nor a constant map , @xmath0 is not semirigid . in order to extend theorem [ cor : additivemaps ] to moufang loops
it would suffice that the identity @xmath388 holds , but this amounts to associativity ( replace @xmath236 by @xmath311 ) , hence to the fact that the moufang loop is a group .
+ let us say that two equivalence relations @xmath4 and @xmath389 on a set @xmath1 are _ orthogonal _ if each class of @xmath4 intersects each class of @xmath389 in at most one element(that is @xmath390 ) and _ strongly orthogonal _ if each class of @xmath4 intersects each class of @xmath389 in exactly one element . according to our introduction ,
a system of three equivalence relations is a _
@xmath77-net _ if and only if these relations are pairwise strongly orthogonal .
[ prop : extending]let @xmath391 be a system of three equivalence relations .
these relations are pairwise orthogonal if and only if @xmath0 is embeddable into a @xmath77-net .
moreover , if @xmath392 is finite , then there is a @xmath77-net of order at most @xmath393 in which @xmath0 is embeddable .
we prove first that the orthogonality conditions are necessary .
next , we prove that they suffice by assuming first that @xmath1 is finite , in which case we apply a result of evans .
if @xmath1 is infinite , we use the diagram method of robinson , a basic technique of mathematical logic .
namely , we define a theory @xmath394 in a first order language ; from the finite case of our proposition , we get that it is consistant , hence by the compactness theorem of first order logic it has some model ; it turns out that such a model is a @xmath77-net extending @xmath0 .
so suppose that @xmath0 is embeddable into a @xmath77-net @xmath395 , let @xmath396 be the image of @xmath1 by an embedding .
since @xmath114 is a @xmath77-net , the @xmath397 s are pairwise orthogonal , hence their restrictions @xmath398 are pairwise orthogonal too .
this is equivalent to the that the @xmath12 s are pairwise orthogonal .
conversely , suppose that these equivalence relations are pairwise orthogonal .
as in the proof of theorem [ theo1 ] , let @xmath194 be the set of equivalence classes of @xmath12 and let @xmath207 be defined by setting @xmath208 .
let @xmath399 , @xmath400 .
extending each @xmath89 to a set @xmath401 of cardinality @xmath402 , the map @xmath4 becomes an embedding of @xmath0 into @xmath403 .
let @xmath404 be defined by setting @xmath405 .
since @xmath406 and @xmath407 are orthogonal , the system @xmath408 is reduced hence , according to theorem [ theo1 ] , the map @xmath409 is an embedding from @xmath408 into @xmath410 .
* assume that @xmath402 is finite .
labelling the elements of @xmath401 by the integers from @xmath311 up to @xmath402 , the range @xmath118 of @xmath1 by @xmath409 appears as a subset of the grid @xmath411 whose elements are labelled by non - negative integers from @xmath311 up to @xmath402 ( corresponding to @xmath412 ) and yields an incomplete latin square . according to evans @xcite , this incomplete latin square extends to a latin square of order @xmath413 .
this latin square provides a @xmath77-net of order @xmath413 in which @xmath0 embeds .
* assume that @xmath402 is infinite .
the definition of the theory @xmath394 goes as follows .
the language consists of three binary predicate symbols @xmath414 , @xmath87 , and infinitely many constant symbols @xmath415 . axioms of @xmath394 are such that every model @xmath416 interpreting predicates and constants yields a @xmath77-net @xmath417 extending @xmath0 through the embedding mapping each @xmath418 to @xmath419 .
for that , we define two sets of axioms ; the first set about @xmath414 , @xmath87 , expresses that @xmath417 will form a @xmath77-net ; the second set that the map @xmath420 is an embedding from @xmath0 into @xmath417 , the axioms needed here are @xmath421 for every @xmath422 distinct in @xmath1 and @xmath423 , resp .
@xmath424 , whenever @xmath425 , resp .
@xmath426 , holds in @xmath0 .
now , if @xmath65 is any finite set of axioms , there is a finite subset @xmath427 of @xmath1 such that no constant symbols @xmath428 with @xmath429 appears among the axioms belonging to @xmath65 . applying the finite case of this proposition to @xmath430 we find a model of @xmath65 in which all constant symbols @xmath431 with @xmath429 are interpreted by some arbitrary element @xmath432 .
since every finite set of axioms has a model , compactness theorem of first order logic ensures that @xmath394 has a model .
a _ triangle _ @xmath394 in @xmath1 consists of three elements @xmath433 ( not necessarily distinct ) of @xmath1 such that @xmath434 , @xmath435 and @xmath436 .
for @xmath437 we say that @xmath438 is @xmath44-_closed _ if any triangle of @xmath44 with two elements in @xmath438 is included in @xmath44 .
denote by @xmath439 the intersection of @xmath44-closed subsets of @xmath44 containing @xmath438 and denote by @xmath440 the set of @xmath441 such that there is a triangle @xmath442 with @xmath443 . for @xmath444 define @xmath445
recursively by @xmath446 and @xmath447 .
[ def : monogenic ] we say that a subset @xmath438 of @xmath44 _ generates _
@xmath44 if @xmath449 . we say that @xmath44 is _ monogenic _ if some subset of @xmath44 with at most two elements generates @xmath44 .
[ examples : semirigid ] if @xmath320 is a cyclic group ( denoted additively ) and @xmath325 , then @xmath1 is monogenic ( indeed , @xmath1 is generated by the pair @xmath450 where @xmath311 generates @xmath320 )
. also for each @xmath451 , the set @xmath452 defined in the introductory section is monogenic .
indeed , the set @xmath453 generates it .
similarly , @xmath454 generates each of @xmath455 and @xmath456 also defined in the introduction .
however , the @xmath457-element set @xmath458 is not monogenic ( see figure [ 2 ] ) .
[ lem : extension ] let @xmath44 be a subset of @xmath1 and @xmath122 , @xmath373 be two homomorphisms of @xmath332 into @xmath0 . if @xmath122 and @xmath373 coincide on a subset @xmath438 of @xmath44 then @xmath122 and @xmath373 coincide on @xmath467 .
it suffices to prove that if @xmath122 and @xmath373 coincide on a subset @xmath438 of @xmath44 they coincide on @xmath440 . for that , let @xmath468 ; our aim is to prove that @xmath469 .
let @xmath470 be a triangle with @xmath443 .
the triangle determinacy applied with @xmath471 , @xmath472 , @xmath473 and @xmath474 yields @xmath475 . according to proposition [ prop : extending ]
, we may suppose that @xmath0 is a @xmath77-net .
more can be said if the associated quasigroup is an additive group , particularly if this group is a subgroup of the additive group of a field .
for each non - empty subset @xmath438 of @xmath1 , let @xmath478 . for @xmath479 , set @xmath480 ( = @xmath481 ) where @xmath482 is the additive subgroup of @xmath19 generated by @xmath483 .
note that the choice of @xmath484 is irrelevant .
[ lem : zede ] if @xmath495 is monogenic then @xmath496 is isomorphic to some system @xmath497 where @xmath498 and @xmath499 is a cyclic group .
furthermore , if @xmath500 contains the vertices of a triangle , then we may assume that the pair @xmath501 ( for some @xmath502 ) generates @xmath500 .
we may suppose that @xmath44 contains the vertices of a triangle , otherwise @xmath44 has at most two elements and the result is trivial . let @xmath503 be a subset of @xmath44 which generates @xmath44 .
then there is some triangle @xmath504 .
clearly @xmath505 and @xmath506 generate @xmath44 .
hence , with no loss of generality , we may suppose that @xmath507 .
thus @xmath508 and @xmath509 according to lemma [ lem : integerenveloppe ] , @xmath44 is included in @xmath486 . according to theorem [ cor : additivemaps ] we may translate @xmath236 and @xmath510 onto @xmath511 and @xmath512 respectively .
if @xmath19 is the additive group of a field , say @xmath513 , multiplication of two elements @xmath514 and @xmath515 of @xmath19 will be denoted by @xmath516 . in this case , a map @xmath322 from @xmath1 into @xmath1 is a _ homothety composed with a translation _ if for some @xmath517 and some _ coefficient _
@xmath518 , @xmath519 for all @xmath520 .
note that if @xmath499 is a cyclic subgroup of @xmath19 and @xmath322 preserves @xmath521 then @xmath522 and @xmath523 ( indeed , since @xmath322 preserves the additive group @xmath118 , @xmath524 ; it follows that @xmath525 for every @xmath526 ; hence , for a generator @xmath527 of @xmath499 , @xmath528 for some @xmath529 ; this yields @xmath522 ) .
we will say that @xmath322 is a _ homothety composed with a translation on @xmath118_. [ prop : monogenic ] suppose that @xmath19 is the additive group of a field .
if a subset @xmath530 is monogenic then every homomorphism of @xmath332 into @xmath0 is the restriction to @xmath44 of the composition of a homothety with a translation .
let @xmath122 be a homomorphism of @xmath332 into @xmath0 and let @xmath503 generates @xmath44 .
set @xmath531 and @xmath532 .
first suppose that @xmath533 .
let @xmath466 be the common value and @xmath534 be the constant map mapping @xmath236 and @xmath510 to @xmath466 .
since @xmath122 and @xmath534 coincide on @xmath438 , lemma [ lem : extension ] ensures that they coincide on @xmath535 .
thus @xmath122 is constant .
now , assume that @xmath536 .
let @xmath537 be such that @xmath538 , set @xmath539 and let @xmath322 be the selfmap of @xmath19 defined by setting @xmath540 for every @xmath541 .
this map is an endomorphism of @xmath542 .
it coincides with @xmath122 on @xmath438 . according to lemma [ lem : extension ] , @xmath122
coincide with @xmath322 on @xmath543 .
[ prop : monogenic2 ] suppose that @xmath19 is @xmath544 or @xmath545 where @xmath546 is prime .
if a subset @xmath547 is monogenic then every endomorphism of @xmath332 is the restriction of the composition of a homothety with a translation on @xmath1 .
we recall that an abelian group @xmath19 is _ torsion - free _ if @xmath548 implies @xmath549 or @xmath550 ) ( @xmath551 ) .
an abelian group @xmath19 is _ divisible _ if for every @xmath552 and every positive integer @xmath26 , the equation @xmath553 has a solution .
torsion - free divisible groups are vector spaces over the field @xmath554 of rational numbers .
also each torsion - free divisible group extends to a field .
[ prop : maintheo ] let @xmath19 be torsion - free and @xmath44 be a monogenic subset of @xmath1 containing @xmath511 .
if each line going through @xmath511 contains only finitely many elements of @xmath44 and if @xmath44 has no center of symmetry then @xmath332 is semirigid .
let @xmath122 be an endomorphism of @xmath332 . according to lemma [ lem : zede ]
we may suppose that @xmath558 . according to proposition [ prop : monogenic2 ] ,
the map @xmath122 is the composition of a homothety with a translation , and there are @xmath555 and @xmath559 such that @xmath560 for all @xmath561 .
we show that @xmath562 and @xmath563 .
suppose for a contradiction @xmath564 .
let us iterate @xmath122 .
for each integer @xmath565 ,
the iterated selfmap @xmath566 satisfies @xmath567 for all @xmath561 . since @xmath568 , obviously @xmath569 for every @xmath565 . hence , since all these elements are multiples of @xmath484 , they all lie on the line going through @xmath511 and @xmath484 .
since , by hypothese this line contains only finitely many elements of @xmath44 , either @xmath570 , in which case @xmath122 is constant , or @xmath571 , in which case @xmath122 is a symmetry whose center is @xmath572 , a case which is excluded by assumption .
hence @xmath562 and thus @xmath573 .
since each line going through @xmath511 contains only finitely many elements of @xmath44 , proceeding in a similar way as above we get @xmath574 .
since @xmath122 is not constant and @xmath511 is not a center of symmetry of @xmath44 , similarly we get @xmath563 . in this case
, @xmath122 is the identity on @xmath44 , thus @xmath0 is semirigid .
there are non - monogenic subsets @xmath44 for which @xmath332 is semirigid .
indeed , let @xmath575 and let @xmath576 be the ones represented figure [ 2 ] .
then @xmath577 is semirigid but @xmath576 is not monogenic ( the semirigidity was checked by computer ) .
we recall that a subset @xmath438 of @xmath27 is _ dense _ if for every @xmath583 in @xmath27 there is some @xmath584 such that @xmath585 .
we also recall that every additive subgroup @xmath158 of @xmath27 is either discrete , in which case @xmath586 for some @xmath587 , or dense .
let @xmath158 be an additive subgroup of @xmath27 containing @xmath544 .
set @xmath588 .
let @xmath373 be an endomorphism of @xmath591 .
in particular , @xmath373 induces a homomorphism of @xmath592 in @xmath0 . since @xmath488 is monogenic then , according to lemma [ prop : monogenic ] ,
the map @xmath593 is the restriction to @xmath488 of the composition of an homothety with a translation , that is @xmath594 for all @xmath595 and some @xmath596 and @xmath517 . since @xmath373 maps @xmath488 into @xmath597
, one would easily see that if @xmath373 is not constant then @xmath563 and @xmath562 and thus @xmath373 is the identity on @xmath488 .
let @xmath598 and set @xmath599 for every @xmath600 .
apply lemma [ lem : additive ] to @xmath601 , @xmath602 and to the restriction of @xmath122 to @xmath44 .
the maps @xmath603 given by lemma [ lem : additive ] coincide and their common value @xmath322 satisfies @xmath604 whenever @xmath605\cap d$ ] .
we claim that @xmath606 for every @xmath607\cap d$ ] .
we observe first that @xmath608 for every rational @xmath609 and @xmath97 such that @xmath610\cap d$ ] .
next we prove that @xmath322 is continuous .
our claim follows .
the continuity of @xmath322 will follows from the fact that @xmath611 for every @xmath607\cap d$ ] ( indeed , from the additivity condition , that means that @xmath322 is @xmath311-lipschitz , hence continuous ) .
so given @xmath607\cap d$ ] , let @xmath612\cap \mathbb q x=(\mathbb q\cap [ 0 , \frac{1}{x}])x$ ] .
observe that @xmath613)h(x))\times\{y_0\}$ ] , thus is a set @xmath614 for a dense subset @xmath615 of a real interval of length @xmath616 .
it is easily checked that no such @xmath617 can be included in @xmath618 if @xmath619 .
now we may conclude that either @xmath373 is constant or the identity .
indeed , if @xmath373 is constant on @xmath488 then , in particular @xmath620 , in which case @xmath621 .
hence , by our claim , @xmath322 is @xmath134 on @xmath622\cap d$ ] .
thus the image of @xmath623 is @xmath484 and @xmath373 is constant . if @xmath373 is not constant on @xmath488 then , as we have seen , @xmath373 is the identity on @xmath488 . in this case
@xmath624 , hence @xmath625 .
thus @xmath626 ; then , by our claim , @xmath322 is the identity on @xmath622\cap d$ ] and @xmath373 is the identity on @xmath623 .
thus , @xmath373 is the identity on @xmath618 , proving that @xmath591 is semirigid . according to zdori s result
, there is a semirigid system of three equivalence on a set of size @xmath58 for each finite @xmath58 distinct from @xmath2 and @xmath3 . if @xmath627 , apply proposition [ maintheo ] .
if @xmath628 , observe ( with the _ axiom of choice _ ) that there are dense additive subgroups of @xmath27 of cardinality @xmath58 . apply proposition [ prop : uncountable ] .
results of this paper have been presented at the ieee 42nd international symposium on multiple - valued logic , ismvl-2012 , may 14 - 16 , victoria , bc , canada , at the special session in honor of ivo g. rosenberg . the authors are pleased to thank pr .
vincent gaudet , michael miller , dan simovici and the organizers of this conference .
they thank lucien haddad for his helpful comments .
they also thank for grant - in - aid for scientific research ( b ) 21300079 , 2009 ( japan ) .
bkkr69a bodnarcuk , v.g . ; kaluznin , l.a . ; kotov , n.n .
; romov , b.a . , galois theory for post algebras i , _ kibernetika _ 3 , 1 - 10 , 1969 ( russian ) .
english translation _ cybernetics and systems analysis _
, 5 , 1969 .
bkkr69b bodnarcuk , v.g .
; kaluznin , l.a . ; kotov , n.n .
; romov , r.a . , galois theory for post algebras ii , _ kibernetika _ 5 , 1 - 9 , 1969 ( russian ) .
english translation _ cybernetics and systems analysis _ , 5,1969
. de groot , j. , groups represented by homeomorphism groups .
* 138 * ( 1959 ) 80102 .
dnes , j. ; keedwell , a. d. , latin squares and their applications . academic press , new york - london , 1974 .
dnes , j ; keedwell , a.d .
, latin squares .
new developments in the theory and applications . with contributions by g. b. belyavskaya , a. e. brouwer , t. evans , k. heinrich , c. c. lindner and d. a. preece . with a foreword by paul erds .
annals of discrete mathematics , 46 .
north - holland publishing co. , amsterdam , 1991 .
xiv+454 pp .
evans , t. , embedding incomplete latin squares .
monthly 67 ( 1960 ) 958961 .
lnger , f. , pschel , r. , relational systems with trivial endomorphisms and polymorphisms , _ j. pure appl .
algebra _ , 32(2):129 - 142 , 1984
. hall , m , jr .
, distinct representatives of subsets . bull .
54 , ( 1948 ) .
miyakawa , m. , pouzet , m. , rosenberg , i. g. and tatsumi h. , semirigid equivalence relations on a finite set , _
j. mult .- valued logic soft comput . _ , 15(4 ) , ( 2009 ) , 395407 .
phelps , k. t. , automorphism - free latin square graphs . discrete math .
31 ( 1980 ) , no .
2 , 193200 . pie68 pierce r.s .
, _ introduction to the theory of abstract algebra _ , holt , rinehalt and winston , new york - london , 1968 + 148 p. strietz h. , " uber erzeugendenmengen endlicher partitionenverbnde , _ studia scien .
math . , _ 12 ( 1977 ) , 1 - 17 .
pouzet , m. ; rosenberg , i.g . ,
general metrics and contracting operations .
graphs and combinatorics ( lyon , 1987 ; montreal , pq , 1988 ) . discrete math . 130
( 1994 ) , no . 1 - 3 , 103 - 169 .
priess - crampe , s. ; ribenboim , r. , generalized ultrametric spaces i , abh . math .
hamburg 66 ( 1996 ) 5573 .
priess - crampe , s. ; ribenboim , r. , generalized ultrametric spaces ii , abh . math .
hamburg 67 ( 1997 ) 1931 .
priess - crampe , s. ; ribenboim , r. , homogeneous ultrametric spaces , j. algebra 186 ( 1996 ) 401435 .
rosenberg , i. g. , strongly rigid relations .
rocky mountain j. math . 3 ( 1973 ) , no .
4 , 631639 . snow , j.w , a constructive approach to the finite congruence lattice representation problem .
algebra universalis 43 ( 2000 ) , no . 2 - 3 , 279293 .
vopnka , p. ; pultr , a. ; hedrln , z. , a rigid relation exists on any set .
carolinae 6 ( 1965 ) 149155 .
zad83 zdori l. , generation of finite partition lattices , _ lectures in universal algebra _ ( proc . colloq . , szeged , 1983 ) , colloq .
jnos bolyai 43 , north - holland , amsterdam , 1986 , 573 - 586 . | a system @xmath0 of equivalence relations on a set @xmath1 is _ semirigid _ if only the identity and constant functions preserve all members of @xmath0 .
we construct semirigid systems of three equivalence relations .
our construction leads to the examples given by zdori in 1983 and to many others and also extends to some infinite cardinalities . as a consequence ,
we show that on every set of at most continuum cardinality distinct from @xmath2 and @xmath3 there exists a semirigid system of three equivalence relations . |
_ the average tranmsission per mode : _ total transmission through a graphene heterojunction can be written as , @xmath74here we have used , angular spacing , @xmath75 , mode spacing @xmath76 and number of modes , @xmath77 . comparing with eq .
[ eq1 ] , we can write , @xmath78transmission through a single @xmath6 junction , where the potential changes smoothely from @xmath14 to @xmath15 over a distance @xmath16 is given by , @xmath79 ignoring the wave - function prefactor , this is valid for moderate gate split distance @xmath16 .
let us consider the @xmath11 for a single split junction and a tilted junction separately .
@xmath81 with gate split . for an abrupt tilted junction , @xmath82 m\label{tilt_scaling}\end{aligned}\ ] ] due to
reduced density of modes at the higher angular region , @xmath83 is scaled with @xmath21 .
therefore , a resistance measurement ( @xmath84 ) will show an increase for a tilted device .
+ _ transmission through dual tilt gpnj device : _ in fig .
[ nano1 ] , we have two such junctions , each of them are tilted .
individual transmissions through the junctions becomes , @xmath85since the tilt angle @xmath21 only modifies the angles of the incoming modes . _
extracting @xmath11 from transport measurement : _ in the experiment @xcite , the junction resistance is extracted from @xmath90/{2},\end{aligned}\ ] ] the above equation eliminates contact and device resistance due to scatterings and leaves out the resistance contribution from the @xmath6 junction only .
theoretically the total resistance @xmath84 can be divided into two parts ( contact and device resistance ) . from eq . 1 ,
@xmath91^{-1}\frac{1}{mt_{av}}\\ & = & [ g_0]^{-1}[\frac{1}{m}+\frac{1-t_{av}}{mt_{av}}]\end{aligned}\]]in presence of a @xmath6 junction with non - unity @xmath11 , the second term can be considered as the junction resistance , @xmath92^{-1}[\frac{1-t_{av}}{mt_{av}}]\end{aligned}\ ] ] while the theoretical @xmath11 is already known ( eq . [ res_t ] ) , the experimental @xmath11 can be found by plugging the value of @xmath93 from measurement in eq .
[ resist_junc ] .
the only unknown value remains is the number of modes at a particular gate voltage .
@xmath94here @xmath95 is the shift of dirac point with gate voltage @xmath96 .
the gate capacitance is calculated from a simple parallel plate capacitor model @xmath97where gate oxide thickness @xmath98 is 100 nm . | we explore the chiral transmission of electrons across graphene heterojunctions for electronic switching using gate geometry alone .
a sequence of gates is used to collimate and orthogonalize the chiral transmission lobes across multiple junctions , resulting in negligible overall current .
the resistance of the device is enhanced by several orders of magnitude by biasing the gates into the bipolar @xmath0 doping regime , as the on state in the near homogeneous @xmath1 regime remains highly conductive .
the mobility is preserved because the switching involves a transmission gap instead of a structural band - gap that would reduce the number of available channels of conduction . under certain conditions
this transmission gap is highly gate tunable , allowing a subthermal turn - on that beats the landauer bound on switching energy limiting present day digital electronics .
the intriguing possibilities of graphene derive from its exceptional electronic and material properties @xcite , in particular its photon - like bandstructure @xcite , ultrahigh mobility @xcite , pseudospin physics and improved 2-d electrostatics @xcite .
its switching ability , however , is compromised by the lack of a band - gap @xcite , while opening a gap structurally kills the available modes for conduction , degrading mobility @xcite .
this begs the question as to whether we can significantly modulate the conductivity of graphene without any structural distortion , thereby preserving its superior mobility and electron - hole symmetry . a way to do
this is to open a transmission gap that simply redirects the electrons , without actually shutting off the density of states .
the dual attributes that help graphene electrons in this regard are its photon like trajectories and chiral tunneling that makes the junction resistance strongly anisotropic , allowing redirection with gate geometry alone . in an earlier paper ,
@xcite we outlined how we can open a transmission gap by a tunnel barrier , angularly injecting the electrons with a quantum point contact ( qpc ) and then selectively eliminating the low incidence angle klein tunneling @xcite modes with a barrier , in that case a patterned antidot or an insulating molecular chain . when the critical angle for total internal reflection is lower than the angle subtended at the qpc by the barrier , electrons are unable to cross over across the junction . the result is a transmission gap that can be collapsed by driving the voltage gradient across the junction towards the homogenous @xmath2 or @xmath3 limit , creating a subthermal turn - on sharper than the landauer binary switching limit of @xmath4 for distinguishability ( @xmath5 for each decade rise in current ) .
beyond proof of concept , that geometry was limited by a paucity of qpc modes and the structural distortions near the barrier that create a larger effective footprint . and room temperature .
the on current degrades slightly compared to homogeneous gapless graphene , but the off current is reduced by several orders of magnitude.,title="fig:",width=326]-3 mm in this paper , we combine a split gated @xmath6 junction to collimate the transverse modes fig . [ nano0](a ) , with recently demonstrated @xcite action of a tilted @xmath6 junction that increases the effective angle of incidence of the electrons .
the conductance at zero temperature can be written as , @xmath7where @xmath8 is the conductance quantum for two spins , @xmath9 is the number of modes , @xmath10 is transmission of individual modes and @xmath11 is the average transmission over all modes . if all modes transmit with equal probability ( @xmath12 ) , the conductance can simply be written as @xmath13 . due to the chiral nature of carriers in graphene , transmission in gpnj
is highly angle ( mode ) dependent making it necessary to work with the average transmission per mode @xmath11 . instead of eliminating the mode
count @xmath9 as does a structural band - gap , we exploit instead the chiral tunneling that makes @xmath11 vanishingly small over a range of energy and controllable with geometry alone ( fig .
[ nano0]b , c ) .
all modes are available for conduction in the on state when the split gates are set to the same polarity and thus retaining high mobility of graphene . _ engineering transmission gap with gate geometry . _
fig .
[ nano0 ] shows two @xmath6 junctions tilted in opposite directions .
each junction exploits chiral tunneling that conserves pseudospin index and maximizes transmission at normal incidence ( klein tunneling ) , especially when they are smooth , i.e. , the @xmath14 to @xmath15 transition occurs over a finite distance @xmath16 .
a tilted junction rotates the transmission lobe accordingly , @xcite , shifting transmissions along opposite directions to make them orthogonal .
the mode - averaged transmissions across the dual junction can be decomposed as below ( see appendix for details ) @xmath17\nonumber \\ & \times & \exp{\biggl[\displaystyle -\pi d\frac{k_{fl}k_{fr}}{k_{fl } + k_{fr } } \sin{(\theta_l \pm \delta)}\sin{(\theta_r)}\biggr]}\nonumber\\ & & \nonumber\\ \displaystyle\frac{1}{t_{eff } } & \approx & \frac{1}{t_1 } + \frac{1}{t_2 } - 1 \\ t_{av}(e_f)&= & \frac{1}{2}\displaystyle \int t_{eff}(\theta)\cos{\theta}d\theta\nonumber\\ & = & [ a\sqrt{k_fd}e^{\pi k_fdsin^2\delta}]^{-1}\end{aligned}\ ] ] which is vanishingly small for moderate doping ( fermi wave - vector , @xmath18 , @xmath19 is a constant , @xmath20 ) , gate split @xmath16 and tilt angle @xmath21 .
the first equation arises from matching pseudospinors across the junction , @xmath22 and @xmath23 denoting components to left and right of a junction ( 1,2 ) .
the tilt angle @xmath21 modifies the incident angle by @xmath24 and the angle of refraction is related to incident angle through snell s law , @xmath25 .
the second equation assumes resistive addition of the junction resistances and ballistic flow in between .
the mode count for an ohmic contacted sample of width @xmath26 is given by @xmath27 .
the resulting total conductance @xmath28 is negligible in the entire @xmath6 junction regime , indicating that the transmission gap ( @xmath29 ) exists if the carrier densities have opposite polarities , @xmath30 where @xmath31 is the gate induced voltage step across the junction .
this is because the high resistance is primarily contributed by the wkb exponential factor which is valid in the @xmath6 regime , whereas the unipolar regime has only the cosine prefactors represeting the wavefunction mismatch @xcite .
vs fermi energy @xmath32 for different doping profiles ( fermi energy @xmath32 and built in potential @xmath31 are indicated on the top band diagram ) .
@xmath11 for the dual tilted gpnj shows a gap ( green line ) , which is termed as transmission gap ( yellow shading ) in this paper.,width=326 ] fig .
[ nano1 ] shows variation of @xmath11 numerically calculated from eq .
[ dev_tr ] as a function of fermi energy ( @xmath32 ) for four different devices and doping profiles .
the orange line shows unit transmission of all modes for a ballistic uniformly doped graphene sheet .
the angular ( mode dependent ) transmission is manifested in a single sharp ( @xmath33 = 0 ) graphene @xmath6 junction and the @xmath11 is suppressed ( blue dots ) .
further suppression is achieved with a split junction ( pink circles ) ( non - zero @xmath33 ) due to high transverse energy ( mode ) filtering .
@xmath11 for the device in fig .
[ nano1](a ) is shown in green , showing a negligible transmission over the bipolar doping regime .
note that both green and pink lines show suppression only in the bipolar doping regime , outside which the exponential scaling in eq.[dev_tr ] is eliminated @xcite .
the minimum current is achieved in @xmath0 regime ( off state ) . over the energy window @xmath34 = [ e_f\,\,\ , e_f - qv_{ds}]$ ] set by the drain voltage @xmath35 ,
@xmath11 varies weakly , so that the off state current at zero temperature for the @xmath0 configuration can be extracted from @xmath36 convolved with the thermal broadening function at finite temperature . for uniformly doped graphene with ballistic transport ,
@xmath37 so that the zero temperature on - off ratio simply becomes , @xmath38^{-1 } \sim a\sqrt{k_fd}(2e^{\displaystyle\pi k_fdsin^2{\delta}})\ ] ] if the biasing is changed all the way from @xmath0 to @xmath39 . fig .
[ nano0](c , pink line ) shows the change in dual tilted gpnj current with gate voltage @xmath40 at room temperature and finite drain bias ( @xmath35 ) , compared with a regular zero bandgap graphene based switch ( black line ) . from the @xmath41 to @xmath39 regime , we see little change in gpnj current on a log scale .
but towards the @xmath0 regime , we see at least three orders of magnitude change when the fermi window remains mostly within the transmission gap .
compared to the blue line , the on current is reduced only slightly , while the off current is reduced by orders of magnitude . the reduction in on current comes due to the fact that the doping is not quite uniform at the on state across the @xmath42 collimator ( maintained at unequal doping to avoid a large voltage swing ) , whereupon the wave - function mismatch leads to lower current than usual . fully ballistic transport assuming an ohmic contacted high quality sample gives us an intrinsic on current in the ma/@xmath43 regime . in this calculation
the gate parameters are @xmath44 , @xmath45 nm , @xmath46v , @xmath47v . with experiment @xcite for a single tilted split junction for several doping conditions .
experiment shows good agreement with the theory confirming the scaling law of tilt , eq .
[ tilt_scaling ] .,width=326 ] critical to the geometric switching is the prominance of angle - dependent chiral transmission across a tilted junction , especially in presence of charge puddles and edge reflection .
fig .
[ nano_expt ] shows the mode averaged transmission extracted ( see method in appendix ) from the measured junction resistance for a single split junction , for varying tilt angles@xcite . for an abrupt tilted junction @xmath48 in the symmetric @xmath6 doping limit and represents an electronic analog of optical malus law .
the reduction in @xmath11 happens due to the angular shift of transmission lobe ( fig .
[ nano0](b ) ) in low angular mode density region @xcite .
the numerically evaluated @xmath11 generalized for a tilted split junction ( solid lines ) agrees with experimental data ( dots ) from all the devices .
this angular dependence persists , for multiple diffusive samples @xcite .
the scaling of @xmath11 in experiment thus confirms the angular shift of the transmission lobes and forms the basis of the proposed device .
the data show a remarkable absence of specular edge scattering , and can be explained by the randomizing effect of roughness . _
modified geometries and impact on subthreshold slope .
_ the aforementioned transmission gap is sensitive to gate parameters .
in particular , making one of the junctions abrupt , using overlapping top and bottom gates , produces additional intricacies in addition to the high on and low off current .
both the geometries in figs .
[ nano0 ] and [ nano2 ] have @xmath6 junctions aimed at filtering out all propagating modes , but in fig .
[ nano2 ] the gate split @xmath49 .
the abruptness of the second junction makes the critical angle more sensitive to gate voltages and the transmission gap in eq .
[ tgap ] needs to be changed .
the first junction limits transmission primarily to the klein tunneling mode @xcite in the off state , while the second junction , tilted at @xmath21 ( fig . [ nano2](a ) ) , increases the effective angle of incidence by the gate tilt angle @xmath21 @xcite ) .
all the electrons are then reflected if the critical angle of the second junction is less than @xmath21 , @xmath50where @xmath51 and @xmath52 are doping concentrations on the two sides of junction 2 .
the resulting transmission vanishes over a range of energies ( following from eq . [ critic ] ) , which can be expressed as @xcite @xmath53 analogous to ref .
@xcite despite being a different ( simpler ) geometry , with the tilt angle @xmath21 replacing the barrier angle @xmath54 .
the tunability of the transmission gap for an abrupt junction bears a direct impact on the rate of change of current with gate voltage . for a semiconductor with fixed bandgap ,
this rate is @xmath55 and limits the energy dissipation in binary switching .
the limit arises from the rate of change in overlap between the band - edge and the fermi - dirac distribution , normally set by the boltzmann tail . in our geometry
however , the transmission gap is created artificially with a gate bias @xmath31 across the gpnj , and can be collapsed by going from heterogeneous ( @xmath0 ) towards the homogenous doping limit ( @xmath39 ) .
such a collapsible transport gap will overlap with the fermi distribution at a higher rate than usual with change in gate bias , leading to a subthermal switching steeper than the landauer limit .
this results a lower gate voltage swing to turn on the device and thus reducing dissipation . _
numerical simulation of quantum flow .
_ to demonstrate carrier trajectories in the proposed device , we numerically solve the non - equilibrium green s function formalism ( negf ) .
the central quantity is the retarded green s function , @xmath56 @xmath57 is the hamiltonian matrix of graphene , described here with a minimal one @xmath58 orbital basis per carbon atom @xmath59 are the self energy matrices for the semi - infinite source and drain leads , assumed to be extensions of the graphene sheet ( i.e. , assuming excellent contacts ) and @xmath60 are the corresponding anti - hermitian parts representing the energy level broadening associated with charge injection and removal .
@xmath61 is the device electrostatic potential .
the current from @xmath62th atom to @xmath63h atom is calculated from @xcite , @xmath64\end{aligned}\]]where the electron correlation function , @xmath65 and in - scattering function , @xmath66 . the source and
drain fermi levels are at @xmath67 and @xmath68 . to see the current distribution in the device
, we apply a small drain bias @xmath35 .
@xmath69 is nonzero only if the @xmath62th atom and @xmath63h atom are neighbors .
the total current at an atomic site can be found by adding all the components , @xmath70 .
fig .
[ nano2 ] ( right column ) shows the local current density .
the @xmath71 on state ( bottom right ) shows little reflection while the @xmath0 off state ( top right ) shows very little current inside the final wedge connected to the drain .
most of the electrons that do not cross the tilted junction are redirected towards the source by the edges .
these electrons , especially the secondary modes , are rejected by the initial collimator and tend to build up in the central wedge .
the build - up of charge increases the local quasi - fermi level @xmath72 until the injection rate at the left junction , set by the transmission rate in eq .
[ dev_tr ] , equals the leakage rate at the right tilted junction , given by the exponentially reduced tail in eq .
[ onoff ] plus additional edge scattering based leakage pathways ( a model was presented in @xcite including a specularity parameter @xmath73 ) . in summary , by manipuilating the angle dependent chiral tunneling of gpnj with patterned gates alone , we can controllably suppress klein tunneling and create a transmission gap as opposed to a structural band - gap .
this is accomplished by combining the angular filtering at a split junction with the experimentally demonstrated chiral tunneling across tilted junctions , such that the transmission lobes across multiple junctions become orthogonal to each other in the off state .
since the on state simply requires changing the polarity of the central gate while sticking with otherwise pristine gapless graphene , the on current stays very high .
furthermore , making the second junction abrupt renders its critical angle and thereby the overall transmission gap highly gate tunable , yielding a subthermal low - voltage turn - on that beats the landauer switching limit .
negf simulations show that the off current is limited by leakage aided by specular edge scattering , and is limited by build - up and blockade of charge in the central angular wedge .
16ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty link:\doibase 10.1103/revmodphys.81.109 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop ( ) , @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1103/physrevb.80.155406 [ * * , ( ) ] @noop ( ) @noop * * , ( ) @noop * * , ( ) @noop _ _ ( , ) |
threefold maximal mixing , ie . tri - maximal mixing , undeniably occupies a special place in the space of all @xmath0 mixings . in some weak basis the two non - commuting mass - matrices @xmath1 , @xmath2 ( @xmath3 ) may simultaneously be written @xcite as ` circulative ' matrices ( ` of degree zero ' ) @xcite : @xmath4 respectively invariant under monomial cyclic permutations ( ` circulation ' matrices @xcite ) of the form : @xmath5 ( @xmath6 , etc.)just as circulant matrices are invariant under simple cyclic permutations @xcite . the mass matrices eq .
1 are diagonalised by the ( eg . circulant ) unitary matrices @xmath7 and
@xmath8 : @xmath9 respectively , leading to threefold maximal mixing : @xmath10 where in all the above and in what follows @xmath11 and @xmath12 represent complex cube - roots of unity and the overhead ` bar ' denotes complex conjugation . after some rephasing of rows and columns the
tri - maximal mixing matrix may be re - written : @xmath13 where the matrix in the parenthesis is identically the character table for the cyclic group @xmath14 ( group elements vs. irreducible representations ) . in
threefold maximal mixing the cp violation parameter @xmath15 is maximal ( @xmath16 ) and if no two neutrinos are degenerate in mass , cp and t violating asymmetries can approach @xmath17 .
observables depend on the sqaures of the moduli of the mixing matrix elements @xcite : @xmath18 in tri - maximal mixing all survival and appearance probabilities are universal ( ie .
flavour independent ) and in particular if two neutrinos are effectively degenerate tri - maximal mixing predicts for the locally averaged survival probability : @xmath19 and appearnce probability : @xmath20 if all three neutrino masses are effectively non - degenerate : @xmath21 .
the super - k @xcite multi - gev data , show a clear @xmath22% suppression of atmospheric @xmath23 for zenith angles @xmath24 @xmath25 0 , as shown in fig . 1a . at the same time
the corresponding distribution for @xmath26 seems to be very largely unaffected , as shown in fig .
the best fit is for ( twofold ) maximal @xmath27 mixing with @xmath28 ev@xmath29 , as shown by the solid / dotted curves in fig . 1 . from the measured suppression we have : @xmath30
whereby the @xmath23 must have a large @xmath31 content , ie .
@xmath32 , or more precisely : @xmath33 where the range quoted corresponds to the @xmath34 error above ( which is largely statistical ) .
the apparent lack of @xmath26 mixing at the atmospheric scale , is supported by independent data from the chooz reactor @xcite ( fig .
2 ) , which rules out large @xmath26 mixing over most of the @xmath35 range allowed in the atmospheric neutrino experiments . taken together ,
the chooz and super - k data indicate that the @xmath31 has no @xmath26 content , ie .
there is a zero ( or near - zero ) in the top right - hand corner of the lepton mixing matrix , @xmath36 @xmath25 @xmath37 .
remarkably , the fritzsch - xing hypothesis @xcite ( published well before the initial chooz data ) predicted just such a zero : @xmath38 the fritzsch - xing ansatz ( eq . 11 ) is readily obtained from theeefold maximal mixing ( eq .
5 ) by the re - definitions : @xmath39 and @xmath40 ( up to phases ) , keeping the @xmath41 tri - maximally mixed . while the _ a priori _ argument for these particular linear combinations @xcite is far from convincing , it is clear that eq .
11 is so far consistent with the atmospheric data : @xmath42 ( cf .
eq . 9 ) ,
while beyond the second threshold : @xmath43 the famous ` bi - maximal ' scheme @xcite is very similar to the fritzsch - xing ansatz and likewise predicts a @xmath44% suppression for the solar data .
taken at face value , the results from the various solar neutrino experiments imply an energy dependent suppression . in particular , taking bp98 fluxes ( and correcting for the nc contribution in super - k ) , the results from homestake and super - k : @xmath45 , lie significantly below the results from the gallium experiments : @xmath46 , as shown in fig . 3 .
mindful of the popularity of the large - angle msw solution and the undenied phenomenological promise of the ` original ' bi - maximal scheme @xcite , we have ourselves proposed @xcite , a phenomenologically viable ( and even phenomenologically favoured ) _ a posteriori _ ` straw - man ' alternative to tri - maximal mixing : @xmath47 which we refer to here as ` optimised ' bi - maximal mixing .
this scheme is of course just one special case of the ` generalised ' bi - naximal hypotheses of altarelli and feruglio @xcite ( and see also ref .
@xcite ) . at the atmospheric scale eq . 14
gives : @xmath48 while beyond the second scale it gives : @xmath49 there is then the added possibility to exploit a large angle msw solution with the base of the ` bathtub ' ( given by the @xmath26 content of the @xmath50 ) given by @xmath51 , as shown in fig .
although clearly the matrix eq .
14 is readily obtained from the tri - maximal mixing matrix eq . 5 by forming linear combinations of the heaviest and lightest mass eigenstates : @xmath52 and @xmath53 ( up to phases ) , with the @xmath50 remaining tri - maximally mixed
, we emphasise that we claim no understanding of why these redefinitions should be necessary .
in general , matter effects deform the mixing matrix and shift the neutrino masses , away from their vacuum values , depending on the local matter density . in the tri - maximal mixing scenario ,
the chooz data require @xmath35 @xmath25 @xmath54 ev@xmath29 ( fig 2 ) , so that matter effects can be very important . for ` intermeduate ' densities
@xcite , the matter mass eigenstates become : @xmath55 and @xmath56 ( up to phases ) while the @xmath31 remains tri - maximally mixed : @xmath57 as evidenced in fig . 4 , the phenomenology of eq .
17 can be almost indistinguishable from that of ` optimised ' bi -
maximal mixing , eq .
14 . beyond the ` matter threshold ' we have for @xmath23 : @xmath58 while for @xmath26 :
@xmath59 for atmospheric neutrinos , account must be taken of the initial flux ratio .
@xmath60 . for @xmath60 @xmath61 @xmath62 , the effective @xmath23 suppression : @xmath63 ( cf . eq .
15 ) while the @xmath26 rate is fully compensated : @xmath64 so that @xmath26 appear not to participate in the mixing . the up / down ratio ( fig . 5 ) measures the effective supression . for energies @xmath65
@xmath66 @xmath67 gev the initial flux ratio @xmath60 @xmath66 @xmath62 and the @xmath26 rate becomes ` over - compensated ' , while , for sufficiently high energies compensation effects vanish as the complete decoupling limit @xmath68 is approached .
the resulting maximum in the up / down ratio for @xmath26 ( fig .
5 ) is describerd as a ` resonance ' by pantaleone @xcite .
as regards the mixing matrix , cp and t violating effects are maximal in tri - maximal mixing , but in spite of this , due to the extreme hierarchy of @xmath35 values involved , for most accessible @xmath69 values , observable asymmetries are expected to be unmeasurably small ( in vacuum ) . in the presence of matter ( or indeed anti - matter ) significant asymmetries can occur , however , ` enhanced ' or ` induced ' by matter effects . thus for example if atmospheric neutrinos are separated @xmath70 , interesting matter induced asymmetries become observable ( fig .
6 ) in tri - maximal mixing .
such effects could be investigated using atmospheric neutrino detectors equipped with magnetic fields @xcite , and/or by using sign - selected beams in long - baseline experiments @xcite .
the vacuum predictions for the solar data in tri - maximal mixing are largely unmodified by matter effects in the sun , as is well known , or , as we have seen , by matter effects in the earth ( eq . 7 vs. eq .
thus we expect @xmath71 in tri - maximal mixing with no energy dependence , as shown in fig . 7 .
( note that also the ` optimsed ' bi - maximal scheme predicts @xmath71 with no energy dependence for @xmath72 outside the ` bathtub ' ) . in fig .
7 , the @xmath73b ( and @xmath74be ) flux , affecting both the super - k and homestake data - points , has been rescaled by an arbitrary factor ( 0.72 ) for comparison to the energy - independent prediction .
given the flux errors ( @xmath75 for @xmath73b @xcite ) , the fit ( fig .
7 ) seems not unreasonable .
8 shows the solar , atmospheric , accelerator and reactor data in overall perspective , within the tri - maximal context .
note that _ dis_-appearance results _ only _ are represented : if the lsnd appearance result @xcite were ever to be confirmed , tri - maximal mixing would be instantly excluded .
the lepton mixing really does look to be either tri - maximal or bi - naximal at this point .
tri - maximal mixing is currently ` squeezed ' in @xmath35 by chooz vs. super - k ( fig . 1 - 2 ) .
for some @xmath72 , bi - maximal mixing ( in particular the ` optimised ' version discussed here ) clearly fits the data better .
tri - maximal mixing remains , however , the ` simplest ' most ` symmetric ' possibility .
9 p. f. harrison and w. g. scott .
b 333 ( 1994 ) 471 .
siam j. matrix anal .
appl . 13 ( 1992 ) 1172 .
s. l. adler .
d59 ( 1999 ) 015012 .
hep - ph/9806518 .
p. f. harrison , d. h. perkins and w. g. scott .
b 349 ( 1995 ) 137 .
b 374 ( 1996 ) 111 .
b 396 ( 1997 ) 186 .
b 458 ( 199 ) 79 . c. walter . eps hep99 ( 1999 ) tampere .
m. appollonio et al .
hep - ex/9907037 .
h. fritzsch and z. xing .
b 372 ( 1996 ) 265 ; b 440 ( 1998 ) 313 .
d. v. ahluwalla . mod .
. lett . a 18 ( 1998 ) 2249 .
a. j. bahz et al .
lett . 81 ( 1998 ) 5730
. v. barger et al .
b 437 ( 1998 ) 107 .
g altarelli and f. feruglio .
b 439 ( 1998 ) 112 . hep - ph/9807353 . c. jarlskog et al .
b 449 ( 1999 ) 240 .
hep - ph/9812282 .
m. b. smy .
hep - ex/9903034 .
j. n. bahcall , s. basu and m. h. pinsonneault .
b 443 ( 1998 ) 1 .
j. pantaleone .
91 ( 1998 ) 5060 .
hep - ph/9810467 .
m. ambrosio et al .
monolith letter of intent .
lngs - loi 20/99 ( 1999 ) .
p. adamson et al .
minos technical design report ( 1998 ) . c. athanassopoulos
lett . 75 ( 1995 ) 2650 ; 81 ( 1998 ) 1774 . | it is argued that data from atmospheric and solar neutrino experiments point strongly to tri - maximal or bi - maximal lepton mixing . while ( ` optimised ' )
bi - maximal mixing gives an excellent _ a posteriori _ fit to the data , tri - maximal mixing is an _ a priori _ hypothesis , which is not excluded , taking account of terrestrial matter effects . |
since the first announcement of a planet orbiting a sun - like star , 51 peg , by @xcite , around 140 stars with planets ( swps ) have been reported .
the main properties of swps have been investigated by many works . among these , spectroscopic observations of most swps based on high - resolution and high signal - to - noise ratio spectra have been carried out .
all of these works indicate a high mean metallicity of swps relative to normal field stars .
are there any other chemical signatures for swps ? there is no conclusive result on this issue , and only a few works have reported abundance anomalies in some elements .
for example , @xcite suggested a high [ c / fe ] ratio for swps as compared with the normal stars from @xcite .
however , they later reported that there is no difference in abundance for c , na , mg , and al elements between swps and normal stars @xcite .
present knowledge on the abundances of 25 elements ( except for li and be ) available in the literature shows no special feature for swps except for the high overall metallicity . for lithium abundance , @xcite claimed that @xmath0 was detected in the atmosphere of hd 82943 , while @xcite did not confirm this detection .
@xcite reported no @xmath1 presence in several lithium - poor stars .
for @xmath2 , @xcite first found several swps with low li abundances , and they related them with the presence of a planetary companion .
@xcite concluded that smaller li abundances are found for swps after correcting for trends with temperature , metallicity , and age .
however , @xcite suggested that there is no significant difference in lithium abundance for swps , and later work by @xcite agreed with ryan s conclusion . with a large sample
, @xcite suggested again that in the temperature range 56005850k , swps show additional depletion of li compared to normal stars .
they also showed that for higher temperatures , the li abundances were similar for swps and for normal stars .
recently , measurements of be abundances by @xcite indicated further evidence on the composition difference between stars with and without planets .
lithium is a special element because it is easily destroyed in stars during even pre main sequence and main sequence evolution , which is not the case for other elements .
therefore , li abundance reflects the mixing history of the star . for swps , li is further related with the accretion of material and the angular momentum evolution of the system , and thus li abundances have been used to distinguish the current hypotheses on planet formation mechanisms . in this respect , it is important to verify whether li abundance anomalies for swps at the temperature range of 56005850k reported by @xcite are real , which will provide new information on the planet formation process and its influence on host stars . in this work
we conduct a parallel abundance analysis of lithium abundances for 16 swps and 20 comparison stars in the same metallicity of @xmath3}}\sim 0.1 $ ] dex at the temperature range of 56005900k in order to investigate whether any anomalies persist .
furthermore , this study is quite interesting for our understanding of the lithium behavior of solar metallicity stars , which show a large scatter at @xmath4 k and which the standard model of stellar evolution can not explain . with a consistent analysis procedure
, this work aims to investigate the effect of planet presence on stellar evolution .
this will provide new information on the chemical evolution of lithium in the galaxy .
swps are selected from a catalog in the temperature range of 56005900k , and their spectra are available from our observations with the 2.16 m telescope of the national astronomical observatories ( xinglong , china ) with a resolving power of 37000 , or from @xcite , who observed with the 1.93 m telescope of the observatoire de haute provence with r@xmath542000 .
@xcite present a database of high resolution spectra of 709 stars covering a large range of atmospheric parameters , which enables us to select a comparison sample of stars spanning the same ranges in temperature , gravity and metallicity as the above sample of planet stars but not having been reported to harbor planets .
that is , both samples cover the atmospheric parameters of @xmath4 k , @xmath6 , and @xmath3 } } > -0.3 $ ] . in the selection ,
strmgren @xmath7 indices from @xcite are adopted and the metallicity is estimated based on the calibrations by @xcite .
along with the initial metallicity , temperature and gravity are derived in the same way as described below .
all spectra have signal - to - noise ratios above 150 pixel@xmath8 at the @xmath96707 line .
the spectra were reduced using standard midas ( for xinglong data ) and iraf ( for elodie data ) routines for order definition , background correction , flat - fielding , extraction of echelle orders , wavelength calibration , and continuum fitting .
the lines suitable for measurement were carefully selected , and the equivalent widths ( ews ) were obtained by the fitting of a gaussian function .
atomic data for lines were the same as those in @xcite . the wavelengths and oscillator strengths of the @xmath96707 line
are taken from @xcite . crccrrrrrrrr star & @xmath10 & @xmath11 & @xmath12 & @xmath3}}$ ] & li & age & ew(1 ) & @xmath13 & ew(2 ) & ew(3 ) & ew(4 ) + hd & k & & @xmath14 & & & gyr & ma & ma & ma & ma & ma + + sun & 5780 & 4.44 & 1.15 & @xmath150.03&@xmath16 & 4.6 & ... & 1.2 & ... & ... & ... + 12661 & 5682 & 4.26 & 1.50 & 0.30 & 1.74 & 9.6&16.6 & 0.8 & ... & ... & ... + 16141 & 5752 & 4.13 & 1.40 & 0.08 & @xmath17 & 7.6 & ... & 1.9 & @xmath18 & ... & ... + 23596 & 5906 & 4.09 & 1.60 & 0.31 & 2.78 & 5.9&86.7 & 1.6 & ... & ... & ... + 33636 & 5817 & 4.41 & 1.30 & @xmath150.03 & 2.29 & 5.8&40.3 & 1.5 & 49.0 & ... & ... + 72659 & 5846 & 4.16 & 1.15 & 0.04 & 2.25 & 6.3&35.4 & 0.6 & ... & ... & ... + 82943 & 5858 & 4.32 & 1.40 & 0.20 & 2.37 & 6.0&44.7 & 2.2&44.1 & ... & ... + 92788 & 5683 & 4.36 & 1.40 & 0.18&@xmath19 & 6.6 & ... & 1.7 & ... & ... & ... + 95128 & 5755 & 4.23 & 1.30 & @xmath150.00 & 1.68 & 9.3 & 10.5 & 2.0&12.5 & ... & 18.0 + 106252 & 5769 & 4.32 & 1.00 & 0.01 & 1.72 & 9.6&13.5 & 2.1 & ... & ... & ... + 134987 & 5735 & 4.27 & 1.15 & 0.24 & @xmath20 & 8.3 & ... & 1.2&@xmath18 & ... & ... + 143761 & 5701 & 4.26 & 1.20 & @xmath150.21 & 1.37 & 10.8&6.2 & 1.1 & 5.9 & ... & 6.0 + 150706 & 5764 & 4.43 & 1.20 & 0.01&2.40 & 7.5 & 53.3 & 1.4 & 55.4 & ... & ... + 187123 & 5717 & 4.26 & 1.15 & 0.11&@xmath21 & 8.8 & ... & 1.8 & ... & ... & ... + 195019 & 5729 & 4.07 & 1.15 & 0.03 & 1.55 & 7.3 & 9.9 & 1.8 & 6.9 & ... & ... + 217014 & 5654 & 4.27 & 1.15 & 0.20&@xmath22 & 10.4 & ... & 0.8&4.4 & ... & ... + + 4307 & 5748 & 3.95 & 1.30 & @xmath150.20 & 2.38 & 6.9&53.9 & 1.9 & 52.3 & ... & ... + 4614 & 5818 & 4.31 & 1.20 & @xmath150.23 & 2.08 & 10.3&21.7 & 1.0 & 26.3 & 24.0 & 21.0 + 9562 & 5786 & 3.99 & 1.20 & 0.18 & 2.51 & 4.6&61.0 & 1.7&61.2 & ... & ... + 15335 & 5797 & 3.93 & 1.40 & @xmath150.15 & 2.55 & 7.0&65.7 & 0.7 & ... & 57.9 & 52.0 + 34411 & 5800 & 4.24 & 1.20 & 0.07 & 2.07 & 7.2&24.3 & 1.1 & 25.7 & 27.4 & ...
+ 39587 & 5833 & 4.41 & 1.40 & @xmath150.00 & 2.85 & 5.8&103.9 & 1.1&101.3 & 105.6 & 95.0 + 52711 & 5778 & 4.31 & 1.30 & @xmath150.10 & 1.84 & 9.2&17.2 & 1.5&17.5 & ... & ...
+ 70110 & 5880 & 3.93 & 1.40 & 0.10 & 2.49 & 3.9&53.2 & 1.9&52.0 & 55.8 & ... + 76151 & 5702 & 4.39 & 1.20 & 0.05 & 1.80 & 6.0&18.2 & 1.6&18.7 & ... & ... + 79028 & 5818 & 4.05 & 1.50 & 0.03 & 2.65 & 7.2&76.1 & 1.4&74.1 & 76.5 & 11.0 + 84737 & 5822 & 4.11 & 1.30 & 0.17 & 2.32 & 5.7&42.9 & 1.5&41.8 & ...
+ 88986 & 5750 & 4.09 & 1.30 & @xmath150.02 & 1.97 & 7.2&23.6 & 1.5&26.2 & ... & ...
+ 109358 & 5742 & 4.28 & 1.20 & @xmath150.26 & 1.59 & 12.0&10.7 & 0.8 & 9.3 & ... & 9.0 + 114710 & 5920 & 4.39 & 1.30 & 0.13 & 2.56 & 2.3 & 63.3 & 1.0&57.8 & ... & ...
+ 115383 & 5918 & 4.19 & 1.30 & 0.19 & 2.78 & 5.2 & 79.2 & 1.3&83.1 & ... & ... + 141004 & 5806 & 4.17 & 1.40 & @xmath150.02 & 1.79 & 8.2 & 14.7 & 1.5&16.0 & ... & 20.0 + 182572 & 5648 & 4.12 & 1.40 & 0.15&@xmath23 & 8.8 & ... & 2.4&@xmath24 & ... & ... + 190406 & 5797 & 4.38 & 1.40 & 0.04 & 2.26 & 5.6 & 39.4 & 1.8 & 37.6 & ... & ... + 193664 & 5795 & 4.39 & 1.20 & @xmath150.11 & 2.27 & 6.1 & 39.9 & 1.2&34.0 & ... & ... + 196755 & 5675 & 3.66 & 1.40 & @xmath150.04 & 1.28 & 3.0 & 6.6 & 1.3&10.8 & ... & ... + [ fig : plot1 ] [ fig : plot2 ] [ fig : plot3 ] [ fig : plot2 ] [ fig : plot3 ] effective temperatures were determined from the strmgren @xmath25 color index @xcite using the calibration of @xcite .
the gravities were determined via _ hipparcos _ parallaxes @xcite , and the microturbulences were obtained by forcing lines with different strengths to give the same abundances . * here
the oscillator strengths of the lines are taken from @xcite . *
the new metallicity derived from spectroscopic analysis was updated , and the procedure of atmospheric parameter determination was iterated for consistency , even though there is no systematic difference between the initial metallicity and the new value , with a deviation of @xmath26 dex .
the internal errors of the atmospheric parameters are around 70k in temperature , 0.1 dex in @xmath11 , 0.1 dex in @xmath3}}$ ] , and 0.3 @xmath14 in microturbulence .
but the absolute uncertainties of the stellar parameters may be slightly larger .
we have found that the photometry - based temperatures in the work are 130 k lower than the spectroscopically derived values in @xcite for swps in common . however , our temperatures are 83 k higher than those in @xcite , who derive temperatures by spectroscopic method for 29 stars in common .
it seems that spectroscopically
derived temperatures have significant uncertainties , and the values depend on stellar model atmosphere , the selection of lines , the adopted atomic data , and the way the temperatures are determined . in addition , as already noted by @xcite , surface gravities determined from the balance of and abundances are systematically higher than those derived from parallaxes ( adopted in our work ) by the order of 0.150.20 dex . however , iron abundances in this work and @xcite are similar since we adopted iron abundances from lines and they forced and lines to give the same iron abundances when they derived gravities . for the comparison stars , there are 15 stars in common with @xcite , and our temperatures are 87k lower than their values .
this is due to different temperature calibrations , and our values , based on the infrared flux method temperatures by @xcite , may be more reliable . actually , the absolute temperature and gravity are not very important , since we are carrying out a differential analysis , and it is crucial that the stellar parameters all be in the same scale between the target and the comparison sample .
the model atmospheres were interpolated from a grid of plane - parallel , lte models provided by @xcite in which convective overshoot is switched off .
the abontest8 program developed by p. magain in the lige group was used to carry out the calculations of theoretical line ews , and abundances were derived by matching the theoretical ews to the observed values .
abundance errors , estimated from the uncertainties of atmospheric parameters and ews , are 0.07 dex for iron abundances , and less than 0.1 dex for li abundances . here
the contribution of the @xmath27 isotope to the @xmath96707 line was assumed to be negligible .
finally , non - lte corrections were applied to the derived li abundances based on the work by @xcite , who studied non - lte formation of the @xmath28 line as a function of effective temperature , gravity , metallicity , and li abundance .
stellar parameters and li and fe abundances for swps and the comparison stars , as well as the ews of the lines and their errors are shown in table 1 .
li abundances as a function of metallicity for all stars in the present survey are shown in fig . 1 ,
in which stars with downward directed arrows have an undetected lithium line , and the symbol size corresponds to stellar age .
the upper limits of the li abundances are estimated by assuming ews of 3.0 m , which is the estimated error in ew from the comparison used in @xcite .
based on table 1 , the scatter of the ew comparison between this work and @xcite , @xcite , and @xcite is around 4.0 m , which indicates an error of @xmath29 massuming the same uncertainties for the ews from the other works .
note that the statistical uncertainties based on the formula by @xcite , presented in table 1 , are smaller .
since experimental errors by different authors during spectrum normalization and ew measurements are not considered in cayrel s ( 1988 ) formula , we believe that an upper limit to the ews of 3.0 m for stars with an undetected li line may be more reasonable .
an interesting result from fig .
1 is that there is a lack of normal stars with @xmath3 } } > + 0.2 $ ] .
this suggestion is supported by large samples of stars in the literature .
there are 22% normal stars with @xmath3 } } > + 0.2 $ ] in @xcite s sample at the temperature range of 56005900k , but note that their temperatures , derived from the fitting of synthetic spectra with observed spectra , are 107k higher than our values for stars in common , and the sample is too small for statistics . with photometry - based temperatures and a larger sample of stars from @xcite , this fraction is reduced to 2% , indicating a low probability of super metal rich stars in the solar neighbourhood .
this is probably due to the fact that most metal rich stars tend to have formed planetary systems , and thus it is difficult to obtain a comparison sample that consists of super metal rich stars not harboring a planet .
this is consistent with the current knowledge that swps have higher mean metallicities than normal stars . the second feature from fig .
1 is that six of the 16 swps versus two of the 20 normal stars have undetected lithium lines .
all of the six swps with undetected li lines are main sequence stars while the two normal stars with depleted li abundances are subgiants , as shown below ( fig . 5 ) .
it seems that main sequence swps in the temperature range of 56005900k destroy their li much more easily than do normal stars .
a kolmogorov - smirnov test is applied to stars with and without planets , and the cumulative distribution functions are shown in fig .
it shows that the maximum value of the absolute difference between the two distribution functions is 0.54 , and the probability for the two samples being the same group is 0.007 .
this is also evident from @xcite , as shown in fig . 3 , in which eight of 16 main sequence swps at @xmath30 k only have upper limits for li abundances , while three of the 14 normal main sequence stars from @xcite are depleted in li .
note that the two comparison samples , from this work and @xcite , are independently selected with six common stars .
thus , this property could be intrinsic , but further investigations with large samples of both swps and normal stars are needed .
if this suggestion that main sequence swps have a higher probability of depleting their lithium abundances is true , it will provide new information on stellar evolution for swps .
we suspect that the stellar evolution history between swps and normal stars could be somewhat different due to the presence of a planet , which appears to affect only li abundances , with no other elements showing any difference .
as suggested by @xcite , the presence of planets or associated circumstellar disks may affect a parent star s initial angular momentum and/or subsequent evolution . specifically , the conservation of angular momentum in the protoplanetary disk may induce increased mixing by causing rotational breaking in the host stars during the pre
main sequence evolution .
this effect , once it happens , will destroy the li completely . considering this suggestion as one possibility , @xcite proposed a second hypothesis : that migration triggers tidal forces and creates a shear instability that leads to a strong depletion of li abundance .
both mechanisms occur during the pre
main sequence evolution , and thus they can be used to explain our result .
further theoretical and observational studies are very desirable to investigate whether the two mechanisms are reasonable .
4 shows li abundances versus temperatures for all stars .
it seems that li abundance generally increases with increasing @xmath10 , and there is no significant gap between swps and normal stars at a given temperature .
5 shows the stellar positions of the two samples of stars in the hr diagram as compared with the evolutionary tracks of @xcite with @xmath3}}\sim + 0.12 $ ] dex .
stellar ages are also estimated from this comparison , and they are presented in table 1 .
the error of the ages due to the uncertainties of @xmath10 , @xmath31 , and @xmath3}}$ ] is about 15% .
it is clear that our two samples of stars are generally located in the same regions of the hr diagram , and during the selection all these stars are classified as main sequence stars with similar parameters .
however , the two stars from the comparison sample showing depleted li abundances , as indicated by downward directed arrows in fig . 5 , are actually subgiants .
other possible subgiant stars are not evolved far away from the main sequence , and there is no difference in li abundance between swps and normal stars .
this observation that subgiants tend to have depleted li abundances reminds us to investigate the stellar positions of swps at @xmath30 k in @xcite .
here we redetermined the temperatures for swps with @xmath30 k from @xcite in the same way as @xcite . in fig . 3 ,
the stellar positions of these swps with new temperatures are plotted against normal stars from @xcite in this temperature range with @xmath3 } } > -0.3 $ ] .
it is clear that a few swps in israelian et al.s ( 2004 ) sample based on the updated temperatures are actually evolved subgiants , and downward directed arrows in fig .
3 indicate stars with undetected @xmath96708 lines .
the fact that evolved subgiants tend to have depleted li abundances is consistent with the knowledge of stellar evolution theory . it has been suggested in @xcite that the li abundances of subgiants can be somewhat depleted due to the dilution proces without complete destruction , as opposed to main sequence stars . in this scenario ,
once a star evolves onto the subgiant branch , it dilutes surface li when convection brings li - poor material from deep layers to the surface .
this dilution is not obvious for slightly evolved subgiants in the present study , and a few subgiants with 1.2 @xmath32 show higher li abundances than main sequence stars .
moreover , there is no difference in li abundance between planet hosts and normal stars on the subgiant branch .
actually , the dilution process mentioned above is difficult to detect , and its effect is masked by the large scatter of li abundances for even slightly evolved subgiants .
it is impossible to investigate how the presence of a planet will affect lithium evolution for evolved subgiants in the present work , since their li abundances have already been depleted due to their low temperatures . as shown in fig . 4
, there is a substantial scatter of li abundances at @xmath33 k. considering the uncertainty in temperature , it is difficult to investigate whether the scatter is real .
there seems to be a tendency that stars in the upper envelope of the li versus @xmath10 diagram seem to be slightly younger than stars with the lowest li abundances in the @xmath34 k range .
however , since there is a lack of young stars in the low temperature range and of old stars at the high temperature edge in our sample , we refrain from drawing any firm conclusions . a larger sample of stars with different ages at a given temperature
is needed to clarify this issue .
the excess of li depletion in swps in the temperature range of @xmath30 k suggested by @xcite is based on the fact that li abundances for swps are lower than for comparison stars .
a closer comparison shows that the average of li abundances for 24 swps at this temperature range in @xcite is @xmath35 , while this value is @xmath36 for the comparison stars from @xcite . in our work
the average li abundance is @xmath37 for swps versus @xmath38 for the comparison stars .
as presented above , this anomaly in li abundance of swps might reflect the effect of planet presence on stellar evolution and can not be used to provide constraints on planet formation .
in addition , this effect is difficult to detect from stars in the subgiant branch or later stages if their li abundances have already been depleted by other mechanisms .
moreover , the deviations of the average li abundance between stars with and without planets are 0.27 dex for @xmath39 k and 0.58 dex for @xmath40k .
it seems that li anomalies concentrate in the range of @xmath40k , and thus they do not relate to the li dip in the hyades , which happens in the temperature range of 63006600k as first found by @xcite . in connection with planet formation mechanism , our result does not support the accretion scenario of planet formation .
theoretically , @xcite predicted quantitative estimates of the main sequence evolution of stellar surface lithium after planet ingestion .
they found that the preservation of @xmath41 occurs in a large mass range of 0.91.3 @xmath32 at solar metallicity .
for swps with a typically super - solar metallicity , the mass range is slightly reduced . according to their theory ,
if a 1 @xmath42 planet were dissolved in the convective envelope of an swp at a sufficiently old age for extra mixing to be inefficient , it would produce an enhancement of 0.3 dex in li abundance , which is unfortunately within the scatter of li abundance at the temperature range we investigated .
but li abundances for stars with @xmath43 k in @xcite do not show such a large scatter , and the similar abundance between swps and the comparison stars argues against the accretion scenario .
furthermore , most swps known so far have giant planets with masses larger than @xmath44 , and if @xmath45 were engulfed the enhancement would reach 0.9 dex , which is easily detected .
therefore , we favor the scenario that the surface composition of swps is not polluted by the accretion of planetary material , and the high mean metallicity of swps is primordial .
we have made a parallel abundance analysis for 16 swps and 20 normal stars at the metallicity of @xmath46 } } < + 0.4 $ ] and the temperature range of @xmath4 k. our results show that there is a higher probability for swps to have depleted li abundances than for comparison stars , which confirms the possible li abundance anomalies for swps at this temperature range reported by @xcite after excluding evolved subgiants .
we proposed that the presence of planets around host stars may affect stellar evolution by inducing additional mixing or shear instability , which leads to a high probability of destroying surface lithium while other elements are not affected .
this effect can only be detected for main sequence swps since other depletion mechanisms would not have started to act on main sequence stars . in view of this , the effect of planet presence should be considered when the large scatters of the galactic lithium in field disk stars at the temperature range of @xmath4 k are investigated .
this work is supported by the national natural science foundation of china under grants 10433010 and 10203002 .
alonso , a. , arribas , s. & martnez - roger , c. 1996 , a&as , 117 , 227 * boesgaard , a.m. , & tripicco , m.j .
1986 , apj , 302 , l49 * carlsson , m. , rutten , r.j . , bruls , j.h.m.j . , & shchukina , n.g .
1994 , a&a , 288 , 860 cayrel , r. 1988 , iau symp . , 132 , 345 chen , y.q . ,
nissen , p.e .
, zhao , g. et al . , 2000 ,
a&as , 141 , 491 chen , y.q . , nissen , p.e . , benoni , t. , & zhao , g. 2001 , , 371 , 943 chen , y.q . , nissen , p.e . , zhao , g. , & asplund m. 2002 , , 390 , 925 dantona , f. , & mazzitelli , 1994 , apjs , 90 , 467 edvardsson , b. , anderson , j. , gustafsson b. , lambert , d.l . , nissen , p.e . , tomkin , j. , 1993 , , 275 , 101 gonzalez , g. , & laws , c. 2000 , , 119 , 390 gonzalez , g. , lawsi , c. , tyagii , s. , & reddy , b.e .
2001 , aj , 285 , 403 gustafsson , b. , karlsson , t. , olsson , e. , edvardsson , b. , & ryde , n. 1999 , a&a , 342 , 426 israelian , g. , santos , n. , mayor , m , & rebolo , r. 2001 , nature , 411 , 163 israelian , g. , santos , n. , mayor , m , & rebolo , r. 2004 , , 414 , 211 king , j.r .
, deliysnnis , c.p .
, hiltgen , d.d .
et al . , 1997 ,
aj , 113 , 1871 kurucz , r.l . , 1995 ,
cdrom , no .
23 lambert , d.l . , heath , j.e .
, & edvardsson , b. , 1991 , mnras , 253 , 610 mandell , a.m. , & ge , j. 2004 , aj , 127 , 1147 mayor , m. , & queloz , d. 1995 , nature , 378 , 357 montalbn , j. , & rebolo , r. , 386 , 1043 nordstrm , b. , mayor , m. , andersen , j. , et al . , 2004 ,
a&a , 418 , 989 olsen e.h . , 1983 , a&as 54 , 55 olsen e.h . , 1993 , a&a 102 , 89 perryman , m.a.c .
, 1997 , the _ hipparcos _ and tycho catalogs ( esa sp-1200 , noordwijk : esa ) prugniel , p. , & soubiran , c. 2001 , a&a , 369 , 1048 reddy , b. , lambert , d.l . , laws , c. , gonzalez , g. , & covery , k. 2002 , , 572 , 1012 ryan , r.g .
2000 , , 316 , l35 santos , n.c . ,
israelian , g. , garca lpex , g.j et al . , 2004 , , 427 , 1085 schuster , b.w . ,
& nissen , p.e .
1989 , a&a , 221 , 65 smith , v.v . ,
lambert , d.l .
, & nissen , p.e .
1998 , , 506 , 405 takeda , y. , & kawanomoto , s. , 2005 , pasj , 57 , 45 valenti , j.a , & fischer , d.a . , 2005 , apjs , 159 , 141 vandenberg , d.a . , swenson , f.j . ,
rogers , f.j . , iglesias , c.a .
, & alexander , d.r .
2000 , , 532 , 430 | we have investigated the abundance anomalies of lithium for stars with planets in the temperature range of 56005900k reported by israelian and coworkers , as compared to 20 normal stars at the same temperature and metallicity ranges .
our result indicates a higher probability of lithium depletion for stars with planets in the main sequence stage .
it seems that stellar photospheric abundances of lithium in stars with planets may be somewhat affected by the presence of planets .
two possible mechanisms are considered to account for the lower li abundances of stars with planets .
one is related to the rotation - induced mixing due to the conservation of angular momentum by the protoplanetary disk , and the other is a shear instability triggered by planet migration .
these results provide new information on stellar evolution and the lithium evolution of the galaxy . |
in nature and technology , a broad variety of active media where pattern selection occurs is governed by kuramoto sivashinsky type equations . in the presence of an imposed advectional transport @xmath0 in the @xmath1-direction the modified kuramoto
sivashinsky equation reads @xmath2 this equation describes two - dimensional large - scale natural thermal convection in a horizontal fluid layer heated from below @xcite and is still valid for a turbulent fluid @xcite , a binary mixture at small lewis number @xcite , a porous layer saturated with a fluid @xcite , etc . in these fluid dynamical systems , except for the turbulent one @xcite , the plates bounding the layer should be nearly thermally insulating ( in comparison to the fluid ) for a large - scale convection to arise . in the problems mentioned , equation governs evolution of temperature perturbations @xmath3 which are nearly uniform along the vertical coordinate @xmath4 and determine fluid currents .
the origin of such a frequent occurrence of equation is its general validity , which may be argued as follows .
basic laws in physics are conservation ones .
this often results in final governing equations having the form @xmath5+\nabla\!\cdot\![\mbox{flux of quantity}]=0 $ ] .
such conservation laws lead to kuramoto sivashinsky type equations .
while the original kuramoto sivashinsky equation has a quadratic nonlinear term ( cf @xcite ) , this term should be replaced by a cubic one for the systems with the sign inversion symmetry of the fields , which is widespread in nature , or for description of a spatiotemporal modulation of an oscillatory mode .
thus the governing equation takes the form . on these grounds , we state
that equation describes pattern formation in a broad variety of physical systems .
we restrict this paper to the case of convection in a porous medium @xcite for the sake of definiteness ; nonetheless , most of our results may be extended in a straightforward manner to the other physical systems mentioned .
is already dimensionless and below we introduce all parameters and variables in appropriate dimensionless forms . in the large - scale ( or long - wavelength ) approximation , which we use ,
the characteristic horizontal scales are assumed to be large against the layer height @xmath6 . in equation
, @xmath7 represents the local supercriticality : @xmath8 is the sum of relative deviations of the heating intensity and of the macroscopic properties of the porous matrix ( porosity , permeability , heat diffusivity , etc . ) from the critical values for the spatially homogeneous case @xcite . for positive spatially uniform @xmath9 , convection sets up , while for negative @xmath9 , all the temperature perturbations decay . in a porous medium @xcite ,
the macroscopic fluid velocity field is @xmath10 where @xmath11 is the stream function amplitude , and the reference frame is such that @xmath12 and @xmath13 are the lower and upper boundaries of the layer , respectively [ ( b ) ] . though the temperature perturbations obey equation for diverse convective systems , function @xmath14 , which determines the relation between the flow pattern and the temperature perturbation , is specific to each case .
notice that @xmath0 is not presented in expression owing to its smallness in comparison to the excited convective currents @xmath15 .
the impact of a weak imposed advective flow on the evolution of temperature perturbations is caused by its symmetry properties : the gross advective flux through the vertical cross - section is @xmath0 , while the convective flow @xmath15 possesses zero gross flux and , therefore , yields a less effective heat transfer along the layer @xcite .
although equation is valid for a large - scale inhomogeneity @xmath7 , which means @xmath16 , one can set such a hierarchy of small parameters , namely @xmath17 , that a frozen random inhomogeneity may be represented by white gaussian noise @xmath18 : @xmath19 where @xmath20 is the disorder intensity and @xmath21 is the mean supercriticality ( i.e. departure from the instability threshold of the disorder - free system ) .
numerical simulation reveals only time - independent solutions to establish in with @xmath22 and such @xmath7 @xcite ; for a small non - zero @xmath0 , stable oscillatory regimes are of low probability by continuity . in the stationary case for @xmath22 the linearized form of equation , i.e.
, @xmath23 is a stationary schrdinger equation for @xmath24 with @xmath21 instead of the state energy and @xmath25 instead of the potential .
therefore , similarly to the case of the schrdinger equation ( see @xcite ) , all the solutions @xmath26 to the stationary linearized equation are spatially localized for arbitrary @xmath21 ; asymptotically , @xmath27 where @xmath28 is the localization exponent .
such a localization can be readily seen for the solution to the nonlinear problem in ( a ) for @xmath29 , @xmath22 , which is a solitary vortex .
for @xmath30 the solitary patterns are still exponentially localized [ ( a ) , @xmath31 .
however , their localization properties change drastically because instead of the second - order linear ode with respect to @xmath24 , equation , one finds a third - order equation : @xmath32 the two symmetric modes @xmath33 and trivial solution @xmath34 ( @xmath35 ) turn into three modes @xmath36 , @xmath37 , of equation with @xmath38 , @xmath39 ( see sample spectrum of @xmath40 in ; cf @xcite for details ) .
specifically , @xmath41-mode is the successor of @xmath42 , @xmath43-mode is the one of the trivial homogeneous mode , and @xmath44-mode is that of @xmath45 .
thus , the upstream flank of the localized pattern is now composed by two modes decaying in the distance from the pattern : @xmath46 where functions @xmath47 and @xmath48 neither grow nor decay over large distances .
the @xmath41-mode , which disappears for @xmath22 , i.e. @xmath49 , decays slowly for a small finite @xmath0 , prevails over the @xmath43-mode decaying rapidly , and , thus , determines the upstream localization properties of the pattern .
the upstream localization length @xmath50 can become remarkably large leading to upstream delocalization of patterns , which can be seen in .
one should keep in mind , that consideration of solitary patterns makes sense where such patterns can be distinguished , i.e. , are sparse enough in space .
this is the case of negative @xmath51 . in , for a sample realization of @xmath18 , one can see that localized patterns can be discriminated for @xmath52 , @xmath53 and the localization properties are very well pronounced for @xmath29 .
here we would like to emphasize the fact of existence of convective currents below the instability threshold of the disorder - free system .
these currents considerably and nontrivially affect transport of a pollutant ( or other passive scalar ) , especially when its molecular diffusivity is small in comparison to the thermal one , which is quite typical in nature ( for instance , at standard conditions the molecular diffusivity of nacl in water is @xmath54 against the heat diffusivity of water which is @xmath55 ) .
transport of a nearly indiffusive passive scalar , quantified by the effective ( or eddy ) diffusivity coefficient , is the object of our research , as a `` substance '' which is essentially influenced by these localized currents and , thus , provides an opportunity to observe manifestation of disorder - induced phenomena discussed in @xcite . in
@xcite we studied the problem for the case of no advection ( @xmath22 ) and calculated ( both numerically and analytically ) the enhancement of the effective diffusivity by disorder - induced currents ; this enhancement is especially strong for low molecular diffusivity deep below the instability threshold of the disorder - free system ( see ) . in this paper
we address the role of an imposed advection in this problem .
the interest to advection is provoked by its dramatic influence on localization properties , i.e. , the upstream delocalization of convective currents that is described above .
we expect this delocalization to result in a giant increase of the effective diffusivity for a nearly indiffusive pollutant and , particularly , in the lowering of the mean supercriticality ( @xmath21 ) value at which the transition from sets of localized convective currents to an almost everywhere intense ` global ' flow occurs .
in this section we describe the transport of a passive pollutant by a steady convective flow ; `` passive '' means that the flow is not influenced by the pollutant .
the assumption of passiveness is practically relevant because ( biologically / chemichally ) significant concentrations of a pollutant can be very small and mechanically negligible .
the flux @xmath56 of the pollutant concentration @xmath57 is @xmath58 where the first term describes the convective transport , the second one represents the molecular diffusion , and @xmath59 is the molecular diffusivity . the establishing time - independent distributions of the pollutant obey @xmath60 [ with account for ] yields a distribution of @xmath57 which is uniform along @xmath4 and is determined by @xmath61 where @xmath62 is the constant pollutant flux along the layer . detailed derivation of equation for @xmath22 can be found in @xcite where it was performed in the spirit of the standard multiscale method ( interested readers can consult , _ e.g. _ , @xcite ) .
remarkably , advection velocity @xmath0 is not presented in the last equation : its direct contribution to convective currents transferring the pollutant is small in comparison to the one of excited convective currents .
instead , it influences the heat transfer and , consequently , excited flows , drastically changing properties of the field @xmath26 .
notice that , for the other convective systems which we mentioned in , the result differs only in the factor ahead of @xmath63 .
thus we come to introducing the effective diffusivity for the system under consideration ( general ideas on the effective diffusivity in systems with irregular currents can be found , e.g. , in @xcite ) .
let us consider the domain @xmath64 $ ] with the imposed concentration difference @xmath65 at the ends .
then the establishing pollutant flux @xmath62 is defined by the integral [ cf ] @xmath66 for a lengthy domain the specific realization of @xmath18 becomes insignificant : @xmath67 hence , @xmath68 which means that @xmath69 can be treated as an effective diffusivity .
the effective diffusivity @xmath70 turns into @xmath59 for vanishing convective flow . for small @xmath59
the regions of the layer where the flow is damped , @xmath71 , make large contribution to the mean value appearing in and diminish @xmath69 , thus , leading to the locking of the spreading of the pollutant .
the disorder strength @xmath20 can be excluded from equations by the appropriate rescaling of parameters and fields . as a consequence
, the results on the effective diffusivity can be comprehensively presented in terms of @xmath72 , @xmath73 , @xmath74 , and @xmath75 : @xmath76 provides calculated dependencies of effective diffusivity @xmath73 on @xmath74 for moderate and small values of molecular diffusivity @xmath72 . concerning these dependencies
the following is worth noticing : * ( a ) * for small @xmath72 a quite sharp transition of effective diffusivity @xmath73 between moderate values and ones comparable with @xmath72 occurs near @xmath77 ( note logarithmic scale of the vertical axis ) , suggesting the transition from an almost everywhere intense ` global ' flow to a set of localized currents to take place . *
( b ) * in the presence of a weak imposed advection , @xmath78 , the transition to ` global ' flow occurs at the value of @xmath74 which is considerably lower than that without advection . *
( c ) * below the instability threshold of the disorder - free system , where only sparse localized currents are excited , the effective diffusion can be significantly enhanced by these currents . *
( d ) * the disorder - induced enhancement of the effective diffusivity is especially drastic in the presence of an imposed advection ; e.g. , for @xmath79 , @xmath80 , the effective diffusivity is increased by one order of magnitude compared to the molecular diffusivity without advection ( @xmath81 ) and by two orders of magnitude for @xmath78 .
the effective diffusivity can be analytically evaluated for a small molecular diffusivity ( @xmath82 ) and sparse domains of excitation of convective currents ( the spacial density of the excitation domains @xmath83 ) . in
@xcite it was evaluated for the case of no advection , @xmath84 where one can use the asymptotic expressions for the density of the excitation domains @xmath85 , @xmath86 and @xmath87 which are valid for @xmath88 .
the latter expression is known from the classical theory of al ( e.g. , see @xcite ) . in the following
we advance the evaluation procedure realized in @xcite in order to account for the asymmetry between up- and downstream localization exponents .
now we calculate the average @xmath89 due to ergodicity , this average over @xmath1 for a given realization of @xmath18 coincides with the average over realizations of @xmath18 at a certain point @xmath90 .
we set the origin of the @xmath1-axis at @xmath90 and find @xmath91 .
when the two nearest to the origin excitation domains are distant and localized near @xmath92 and @xmath93 ( see ) , @xmath94 where @xmath95 and @xmath96 characterize the amplitude of temperature perturbation modes excited around @xmath97 and @xmath98 , respectively . for small @xmath72 and density @xmath85 , the contribution of the excitation domains to @xmath99 is negligible against that of the extensive regions where flow is weak , but @xmath100 is still larger than @xmath72 .
therefore , one may be not very subtle with `` cores '' of excitation domains and may utilize expression even for small @xmath101 : @xmath102 where @xmath103 [ @xmath104 is the density of the probability to observe the nearest right [ left ] excitation domain at @xmath105 [ @xmath106 . for probability distribution @xmath107 , one finds @xmath108 , i.e. , @xmath109 .
hence , @xmath110 , and probability density @xmath111 . as regards averaging over @xmath112 ,
it is important that the multiplication of @xmath112 by factor @xmath113 is effectively equivalent to the shift of the excitation domain by @xmath114 , which is insignificant for @xmath115 in the limit case that we consider .
hence , one can assume @xmath116 ( the topological difference between different combinations of signs of @xmath112 is not to be neglected ) and rewrite equation as @xmath117 .
\nonumber\end{aligned}\ ] ] for @xmath118 and @xmath43-mode dominating over @xmath41-mode ( that is the case in ) , the last formula yields @xmath119 here we assume that advection is weak and suppresses thermal convection in a negligible fraction of the excitation centers and the asymptotic expression is still valid .
noticeable difference between equation for @xmath81 , i.e. , @xmath120 , and equation is actually insignificant up to our approximations , because moderate number @xmath121 risen to small power @xmath122 , @xmath123 which is the ratio of these equations , approximately equals 1 . is simpler than that we consider here and admits analytical evaluation of integrals with a fewer number of approximations . ] for instance , in , the curves given by analytic expressions and with @xmath120 are visually undistinguishable . for small finite @xmath75 , smallness of @xmath43 in equation
gives rise to a significant enhancement of effective diffusivity @xmath73 , which is in agreement with the results of numerical simulation presented in .
shows that for @xmath78 the effective diffusivity in the presence of advection is always stronger than without it ; the larger the difference between the effective and the molecular diffusivity the stronger advectional enhancement of the effective diffusivity is . for instance , for @xmath124 , @xmath80 the effective diffusivity in the presence of advection @xmath78 is by factor 10 larger than without advection , and this factor grows as @xmath72 decreases .
noteworthy , for @xmath78 expression provides slightly overestimated value of the effective diffusivity while for @xmath81 the analytical estimation is accurate .
the inaccuracy appears because in our analytical theory we ignore three factors : ( a ) decrease of the spatial density of the excitation centers owing to advectional suppression ( washing - out ) of weak excitation centers ; ( b ) for small @xmath75 the rapidly decaying upstream @xmath41-mode is significant because of the smallness of the slowly decaying @xmath43-mode ; and ( c ) as the advection strengthens , currents in some excitation domains disappear via a hopf bifurcation @xcite and thus there is non - zero probability to observe oscillatory flows for small @xmath75 even though there is no stable time - dependent solutions for @xmath81 .
unfortunately , inaccuracies caused by these three assumptions can not be minimized simultaneously : the first and third assumptions require @xmath125 , while the second one needs @xmath75 to be small but finite .
oscillatory localized patterns discovered in the dynamic system for non - zero @xmath75 ( ( a ) ; see @xcite for details ) are statistically improbable and rare when @xmath75 is small .
notably , their relative contribution to the effective diffusivity is much larger than their fraction among the excited localized patterns . in ( b )
one can see the soaring of the effective diffusivity along a finite region as a localized pattern turns oscillatory ( @xmath126 ) before disappearing ( @xmath127 ) .
( a ) reveals the origin of this soaring : the oscillatory pattern is not so well localized as the time - independent one .
indeed , the localization properties of the oscillatory pattern of frequency @xmath128 are determined by the following linearization of equation : @xmath129 in contrast to , this is already a 4th - order differential equation , which yields four localization exponents .
the newly appeared 4th mode possesses @xmath130 , i.e. , decays slowly , and contributes to the downstream flank of localized patterns ( evidence of these facts is beyond the scope of this paper and will be presented elsewhere ) . as well as
the @xmath43-mode results in upstream delocalization of time - independent patterns , the new mode leads to downstream delocalization of oscillatory patterns , which appear to be weakly localized both up- and downstream , as one sees this in ( a ) .
nevertheless , owing to the smallness of the fraction of the oscillatory patterns among all the localized patterns at small @xmath75 , their contribution to the effective diffusivity over large domains is still negligible .
this is additionally confirmed by the accuracy of our analytical theory disregarding oscillatory currents [ equation ] .
meanwhile , the analytical theory accounting for the oscillatory patterns should involve the distribution of frequencies of excited patterns , which are to be determined only from the nonlinear problem : this is not an analytically solvable problem .
we have studied the transport of a pollutant in a horizontal fluid layer by spatially localized two - dimensional thermoconvective currents appearing under frozen parametric disorder in the presence of an imposed longitudinal advection .
though we have considered the specific physical system , a horizontal porous layer saturated with a fluid and confined between two nearly thermally insulating plates , our results can be in a straightforward manner extended to a broad variety of fluid dynamical systems ( like ones studied in @xcite ) .
we have calculated numerically the dependence of the effective diffusivity on the molecular one and the mean supercriticality for a non - zero advection strength ( see ) .
the results reveal that advectional delocalization of convective currents greatly assists transfer of a nearly indiffusive pollutant ( @xmath131 ) below the instability threshold of the disorder - free system : the effective diffusivity can become by several orders of magnitude larger in comparison to that without advection . the analytical theory focusing on advectional delocalization of localized current patterns yields results which are in a fair agreement with the results of numerical simulation .
this correspondence confirms our treatment of importance of disorder - induced patterns and their localization properties in active / dissipavite media , which provoked works @xcite . | frozen parametric disorder can lead to appearance of sets of localized convective currents in an otherwise stable ( quiescent ) fluid layer heated from below .
these currents significantly influence the transport of an admixture ( or any other passive scalar ) along the layer .
when the molecular diffusivity of the admixture is small in comparison to the thermal one , which is quite typical in nature , disorder can enhance the effective ( eddy ) diffusivity by several orders of magnitude in comparison to the molecular diffusivity .
in this paper we study the effect of an imposed longitudinal advection on delocalization of convective currents , both numerically and analytically ; and report subsequent drastic boost of the effective diffusivity for weak advection .
_ special issue _ : article preparation , iop journals + disorder in operation conditions of a dynamic system is known to be able to play not only trivial destructive role , distorting the system behavior , but also constructive one , inducing certain degree of order and leading to various non - trivial effects : anderson localization @xcite , stochastic @xcite and coherence resonances @xcite , noise - induced synchronization @xcite , etc .
one of the most distinguished and fair effects is the anderson localization ( al ) , which is the localization of states in spatially extended linear systems subject to a frozen parametric disorder ( random spacial inhomogeneity of parameters ) .
al was first discovered and discussed for quantum systems @xcite .
later on , investigations were extended to diverse branches of semiclassical and classical physics : wave optics @xcite , acoustics @xcite , etc .
the phenomenon was comprehensively studied and well understood mathematically for the schrdinger equation and related mathematical models @xcite .
the role of nonlinearity in these models was addressed in literature as well ( e.g. , @xcite ) .
while extensively studied for conservative media , the localization phenomenon did not receive comparable attention for active / dissipative ones , like in problems of thermal convection or reaction - diffusion .
the main reason is that the physical interpretation of formal solutions to the schrdinger equation is essentially different from that of governing equations for active / dissipative media ; therefore , the theory of al may be extended to the latter only under certain strong restrictions ( see @xcite for reference ) .
nevertheless , effects similar to al can be observed in fluid dynamical systems @xcite . in @xcite we addressed the problem where localized thermoconvective currents excited in a horizontal porous layer under frozen parametric disorder ( spacial inhomogeneity of the macroscopic permeability , the heat diffusivity , etc . )
drastically influence the process of transport of a passive scalar ( e.g. , a pollutant ) along the layer .
below the threshold of instability of the disorder - free system , the effective diffusivity quantifying this transport has been found to be faithfully determined by localization properties of patterns .
meanwhile , in @xcite these properties have been revealed to be greatly affected by a weak imposed longitudinal advection .
hence , one can expect a weak advection to lead to a significant enhancement of the effective diffusivity of a nearly indiffusive pollutant .
treatment of this effect is the subject of the present paper .
the paper is organized as follows . in
we formulate the specific physical problem that we deal with and introduce the relevant mathematical model ; in particular , we discuss disorder - induced excitation of localized currents below the instability threshold of the disorder - free system and advectional delocalization of these patterns . presents the results of a numerical simulation and calculation of the effective diffusivity . in
we develop an analytical theory for the effective diffusivity in the presence of an imposed advection . ends the paper with conclusions . |
student resistance to the evaluation of @xmath0 as the real number @xmath1 ( henceforth referred to as the _ unital evaluation _ ) has been widely discussed in the mathematics education literature .
it has been suggested that the source of such resistance lies in a psychological predisposition in favor of thinking of @xmath0 as a process , or iterated procedure , rather than the final outcome , see for instance d. tall s papers @xcite , @xcite , @xcite ( see also @xcite for another approach ) .
we propose an alternative model to explain such resistance , in the framework of non - standard analysis . from this point of view
, the resistance is directed against an unspoken and unacknowledged application of the standard part function `` st '' ( see section [ ten ] , item [ 103 ] ) , namely the stripping away of a ghost of an infinitesimal , to echo george berkeley @xcite , implicit in unital evaluation : @xmath5{9tail.ps}}}\end{pmatrix } = 1.\ ] ] what is the significance of such fine ( indeed , infinitesimal ) districtions between @xmath0 and @xmath1 ? as this text is addressed to a wide audience , it may be worth recalling that _ dividing _ by the difference @xmath6 will only be possible if the latter is nonzero , a matter of immediate import for studying rates of change and calculating derivatives ( see section [ zoom ] ) .
the hyperreal approach to a certain extent vindicates the student resistance to education professionals toeing the standard line on unital evaluation : so long as the number system has not been specified , the students hunch that @xmath0 can fall infinitesimally short of @xmath1 , can be justified in a mathematically rigorous fashion . in section [ two ] ,
based on a formula for a geometric sum , we summarize the standard argument in favor of unital evaluation . in section [ three ] , we summarize the standard resistance to the latter . in section [ four ] , we construct a hyperreal decimal in @xmath7 , represented by a string of @xmath4-infinitely many repeated 9s , and identify its lightstone representation . in section [ five ] , we square the strict inequality with standard reality , by means of the standard part function . in section
[ six ] , we represent the hyperreal graphically by means of an infinite - magnification microscope , already exploited for pedagogical purposes by keisler @xcite and tall @xcite . in section [ zoom ] , we exploit the hyperreal to calculate @xmath8 . in section
[ eight ] , we develop an applied - mathematical model of a hypercalculator so as to explain a familiar phenomenon of a calculator returning a string of 9s in place of an integer . in section
[ nine ] , we examine the cognitive concept of _ generic limit _ of cornu and tall in mathematics education , in relation to a hyperreal approach to limits . in section
[ flat ] , we set the situation in perspective , with e. a. abbott .
section [ ten ] is a technical appendix containing basic material on non - standard calculus .
the historical section [ history ] contains an examination of the views of courant , lakatos , and e. bishop .
section [ proposal ] contains a 10-step proposal concerning the problem of unital evaluation .
evaluating the formula @xmath9 at @xmath10 , we obtain @xmath11 or alternatively @xmath12 multiplying by @xmath13 , we obtain @xmath14 for every @xmath15 . as @xmath16 increases without bound , the formula @xmath17 becomes @xmath18 or does it ?
when i tried this one on my teenage daughter , she remained unconvinced . she felt that the number @xmath0 is smaller than @xmath1 .
after all , just look at it !
there is something missing before you reach @xmath1 .
i then proceeded to give a number of arguments .
apologetic mumbo - jumbo about the alleged `` non - unicity of decimal representation '' fell on deaf ears .
the one that seemed to work best was the following variety of the old - fashioned `` because i told you so '' argument : factor out a @xmath19 : @xmath20 to obtain @xmath21 and `` everybody knows '' that the number @xmath22 is exactly `` a third '' _ on the nose_. q.e.d .
this worked for a few minutes , but then the validity of was called into question , as well .
then i finally broke down . in abraham robinson s theory of hyperreal analysis @xcite , there is a notion of an infinite hyperinteger ( see section [ ten ] , item [ 108 ] ) .
h. jerome keisler @xcite took to denoting such an entity by the symbol @xmath23 possibly because its inverse is an infinitesimal @xmath24 typically appearing in the denominator of the familiar definition of derivative ( it most decidedly does _ not _ stand for `` howard '' ) .
taking infinitely many terms in formula is interpreted as replacing @xmath15 by an infinite hyperinteger @xmath25 the transfer principle ( see section [ ten ] , item [ 101 ] ) applied to then yields @xmath26 where the infinitesimal quantity @xmath27 is nonzero : @xmath28 therefore we obtain the strict nonstandard inequality @xmath29 and my teenager was right all along .
note that hyperreal extended decimal expansions were treated by a. lightstone in @xcite . in his notation , the hyperreal appearing in the left - hand - side of appears as the extended decimal @xmath30 with the hat `` @xmath31 '' indicating the @xmath4-th decimal place , where the _ last _ repeated digit @xmath3 occurs .
we have employed the underbrace notation as in rather than lightstone s semicolon , as it parallels the finite case more closely , and seems more intuitive .
an alternative construction of a strict inequality @xmath32 may be found in @xcite , however in a number system which is not a field .
to obtain a real number in place of the hyperreal @xmath34 , we apply the standard part function `` @xmath35 '' ( see section [ ten ] , item [ 103 ] ) : @xmath36 to elaborate further , one could make the following remark . even in standard analysis ,
the expression @xmath0 is only shorthand for the * limit * of the finite expression when the standard integer @xmath16 increases without bound . from the hyperreal viewpoint , `` taking the limit '' is accomplished in two steps : 1 .
evaluating the expression at an infinite hyperinteger @xmath4 , and then 2 . taking the standard part .
the two steps form the non - standard definition of limit ( see section [ ten ] , item [ 1010 ] ) . now
the first step ( evaluating at @xmath4 ) produces a hyperreal number dependent on @xmath4 ( in all cases it will be strictly less than @xmath1 ) .
the second step will strip away the infinitesimal part and produce the standard real number @xmath1 in the _ cluster _ ( see section [ ten ] , item [ 104 ] ) of points infinitely close to it .
[ 51b ] the existence of more than one infinite hyperreal is not only a requirement to have a field , but is actually extremely useful . for example , using the natural hypperreal extension @xmath37 of @xmath38 ( see section [ ten ] , item [ 101 ] ) , it is possible to write down a pointwise definition of uniform continuity of a function @xmath38 ( see below ) .
such a definition considerably reduces the quantifier complexity of the standard definition . to elaborate ,
note that the standard definition of uniform continuity of a real function @xmath38 can be said to be _ global _ rather than _ local _
( i.e. pointwise ) , in the sense that , unlike ordinary continuity , uniform continuity can not be defined as a pointwise property of @xmath38 .
meanwhile , in the framework of robinson s theory , it is possible to give a definition of uniform continuity of the real function @xmath38 in terms of its natural hyperreal extension , denoted @xmath39 in such a way that the definition is local , in the sense of depending only on each pointwise cluster ( see section [ ten ] , item [ 104 ] ) in the domain of @xmath37 .
thus , @xmath38 is uniformly continuous on @xmath40 if the following condition is satisfied : @xmath41 here @xmath42 stands for the relation of being infinitely close . the condition must be satisfied at the infinite hyperreals @xmath43 , in addition to the finite ones .
this addition is what distinguishes uniform continuity from ordinary continuity .
the symbol `` @xmath44 '' is employed in standard real analysis to define a formal completion of the real line @xmath40 , namely @xmath45 ( sometimes a formal point `` @xmath46 '' is added , as well ) .
such a formal device is helpful in simplifying the statements of certain theorems ( which would otherwise have a number of subcases ) .
the symbol is used in a _
different _ sense in topology and projective geometry , where the thom compactification @xmath47 of @xmath40 is a circle : @xmath48 we have refrained from using the symbol `` @xmath44 '' to denote an infinite hyperreal , as in @xmath49 even though the symbol `` @xmath44 '' does convey the idea of the infinite more effectively than the symbol `` @xmath4 '' that we have used .
the reason is so as to avoid the risk of creating a false impression of the uniqueness of an infinite point ( as in or above ) , in the field @xmath50 ( see section [ ten ] , item [ 102 ] ) .
we represent the hyperreal @xmath51 visually by means of an infinite - resolution microscope already exploited for pedagogical purposes by keisler @xcite .
the hyperreal @xmath34 appearing in the diagram of figure [ micro ] illustrates graphically the strict hyperreal inequality @xmath52 where , as we already mentioned , the symbol @xmath4 is exploited to denote a fixed infinite robinson hyperinteger ( see section [ ten ] , item [ 108 ] ) .
to calculate the slope of the tangent line to the curve @xmath53 at @xmath54 , we first compute the ratio @xmath55 where we have deleted the underbrace @xmath56 and also shortened the symbol @xmath0 to @xmath57 , so as to lighten the notation . at @xmath54,height=288 ]
therefore the slope is @xmath58 note that @xmath59 in lightstone s notation , where the digit `` 1 '' appears in the @xmath60-decimal place , as illustrated in figure [ slope ] .
everyone who has ever held an electronic calculator is familiar with the curious phenomenon of it sometimes returning the value @xmath61 in place of the expected @xmath62 .
for instance , a calculator programmed to apply newton s method to find the zero of a function , may return the @xmath63 value as the unique zero of the function @xmath64 developing a model to account for such a phenomenon is complicated by the variety of the degree of precision displayed , as well as the greater precision typically available internally than that displayed on the lcd . to simplify matters
, we will consider an idealized model , called a _
hypercalculator _ , of a theoretical calculator that applies newton s method precisely @xmath4 times , where @xmath4 is a fixed infinite hyperinteger , as discussed in the previous section .
let @xmath38 be a concave increasing differentiable function with domain an open interval @xmath65 and vanishing at its midpoint .
then the hypercalculator applied to @xmath38 will return a hyperreal decimal @xmath0 with an initial segment consisting of an unbounded number of repeated @xmath3 s .
assume for simplicity that @xmath66 .
we have @xmath67 by the mean value theorem , there is a point @xmath68 such that @xmath69 where @xmath70 , or @xmath71 since @xmath38 is concave , its derivative @xmath72 is decreasing , hence @xmath73 thus @xmath74 .
inductively , the point @xmath75 satisfies @xmath76 for all @xmath16 . by the transfer principle
( see section [ ten ] , item [ 101 ] ) , the hyperreal @xmath77 satisfies a strict inequality @xmath78 as well ( cf . ) .
hence the hypercalculator returns a value strictly smaller than @xmath1 yet infinitely close to @xmath1 , proving the theorem .
the precise meaning of the finite expression @xmath79 is that the repeated digit 9 occurs precisely @xmath16 times .
the standard non - terminating decimal @xmath80 as it is traditionally written , is said to have an unbounded number of repeated digits 9 , but the expression infinitely many 9 s " is only a figure of speech , as infinity " is not a number in standard analysis , in the sense that , whenever a precise meaning is attributed to the phrase infinitely many 9 s " , it is almost invariably in terms of _
d. tall writes in @xcite as follows : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ [ ... ] the infinite decimal @xmath81 is intended to signal the _ limit _ of the sequence @xmath82 @xmath83 @xmath84 , ... which is @xmath1 , but in practice it is ofter imagined as a limiting process which never quite reaches @xmath1 . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ tall @xcite describes a concept in cognitive theory he calls a _
generic limit concept _ in the following terms : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ [ ... ] if a quantity repeatedly gets smaller and smaller and smaller without ever being zero , then the limiting object is naturally conceptualised as an extremely small quantity that is not zero ( cornu @xcite ) .
infinitesimal quantities are natural products of the human imagination derived through combining potentially infinite repetition and the recognition of its repeating properties . _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ ( see @xcite for the related notion of _ procept _ ) . a technical realisation of the cognitive concept of _ generic limit _
is achieved in the hyperreal line ( see section [ ten ] , item [ 102 ] ) as follows .
given an infinite hyperinteger @xmath4 , consider the hyperreal repeated decimal where the repeated digit @xmath3 occurs precisely @xmath4 times , in the sense routinely used in non - standard calculus , for example , when one partitions a compact interval into @xmath4 parts in the proof of the extreme value theorem ( see section [ ten ] , item [ 109 ] ) .
such a number can be denoted suggestively by @xmath34 with an underbrace indicating that the digit @xmath3 occurs @xmath4 times , resulting in a strict inequality @xmath85 ( with the underbrace indicating that we are not referring to the standard real ) . in lightstone
s notation , this hyperreal would be expressed by the hyperdecimal @xmath86 the last digit 9 occurring in the @xmath4-th position .
such a hyperreal appears to be the mathematical counterpart of the cognitive concept of _ generic limit_.
as far as standard limits are concerned , given the sequence @xmath87 , @xmath88 , @xmath89 , etc .
, from the hyperreal viewpoint we have @xmath90 where st " is the standard part function which strips away " the infinitesimal part , and outputs the standard real in the cluster of the hyperreal @xmath91 ( see section [ ten ] , item [ 104 ] ) .
what may be bothering the students is this unacknowledged application of the standard part function , resulting in a loss of an infinitesimal .
we have , in fact , been looking at the problem `` from above '' , in the context of non - standard analysis .
perhaps a helpful parallel is provided by the famous animated film _ flatland _ ( cf .
@xcite ) , where the two dimensional creatures are unable to conceive of what we think of as the sphere in 3-space , due to their dimension limitation .
similarly , one can conceive of the difficulty in the understanding of the unital evaluation of @xmath0 , as due to the limitation of the standard real vision .
the notion of an infinitesimal appeals to intuition ( see r. courant s comment in section [ history ] ) and would not go away inspite of what is , by now , over a century of @xmath92-ideology ( see e. bishop s comment in section [ history ] ) .
highschool students are exposed to the thorny @xmath0 issue before they are exposed to any rigorous notion of a _
real number_. they are not aware of fine differences between rational numbers , algebraic numbers , real numbers , hyperreal numbers .
a related point is make by keisler in his textbook @xcite , when he points out that `` we have no way of knowing what the line in physical space looks like '' .
most students ( perhaps all ) initially believe that the mysterious number with `` infinitely many '' repeated digits 9 falls short of the value @xmath1 . if an education professional claims that the students are making a mistake , might he in fact be making a pedagogical error ?
in this section we present some illustrative terms and facts from non - standard calculus @xcite .
the relation of being infinitely close is denoted by the symbol @xmath42 .
thus , @xmath93 if and only if @xmath94 is infinitesimal . the _ extension principle _ of non - standard calculus states that every real function @xmath38 has a hyperreal extension , denoted @xmath37 and called the natural extension of @xmath38 .
the _ transfer principle _ of non - standard calculus asserts that every real statement true for @xmath38 , is true also for @xmath37 .
for example , if @xmath95 for every real @xmath43 in its domain @xmath96 , then @xmath97 for every hyperreal @xmath43 in its domain @xmath98 . note that if the interval @xmath96 is unbounded , then @xmath98 necessarily contains infinite hyperreals .
we will typically drop the star @xmath99 so as not to overburden the notation .
internal set is the key tool in formulating the transfer principle , which concerns the logical relation between the properties of the real numbers @xmath40 , and the properties of a larger field denoted @xmath100 called the _
hyperreal line_. the field @xmath50 includes , in particular , infinitesimal ( infinitely small " ) numbers , providing a rigorous mathematical realisation of a project initiated by leibniz . roughly speaking ,
the idea is to express analysis over @xmath40 in a suitable language of mathematical logic , and then point out that this language applies equally well to @xmath50 .
this turns out to be possible because at the set - theoretic level , the propositions in such a language are interpreted to apply only to internal sets rather than to all sets .
note that the term language " is used in a loose sense in the above .
a more precise term is _ theory in first - order logic_. internal sets include natural extension of standard sets .
the standard part function
st " is the key ingredient in abraham robinson s resolution of the paradox of leibniz s definition of the derivative as the ratio of two infinitesimals @xmath101 the standard part function associates to a finite hyperreal number @xmath43 , the standard real @xmath102 infinitely close to it , so that we can write @xmath103 in other words , `` st '' strips away the infinitesimal part to produce the standard real in the cluster . the standard part function
st " is not defined by an internal set ( see item [ 102 ] above ) in robinson s theory .
each standard real is accompanied by a cluster of hyperreals infinitely close to it .
the standard part function collapses the entire cluster back to the standard real contained in it .
the cluster of the real number @xmath104 consists precisely of all the infinitesimals .
every infinite hyperreal decomposes as a triple sum @xmath105 where @xmath4 is a hyperinteger , @xmath106 is a real number in @xmath7 , and @xmath107 is infinitesimal .
varying @xmath107 over all infinitesimals , one obtains the cluster of @xmath108 . to define the real derivative of a real function @xmath38 in this approach ,
one no longer needs an infinite limiting process as in standard calculus .
instead , one sets @xmath109 where @xmath107 is infinitesimal , yielding the standard real number in the cluster of the hyperreal argument of `` st '' ( the derivative exists if and only if the value is independent of the choice of the infinitesimal ) .
the addition of `` st '' to the formula resolves the centuries - old paradox famously criticized by george berkeley @xcite ( in terms of the _ ghosts of departed quantities _ , cf .
* chapter 6 ) ) , and provides a rigorous basis for the calculus .
a function @xmath38 is continuous at @xmath43 if the following condition is satisfied : @xmath110 implies @xmath111 .
a function @xmath38 is uniformly continuous on @xmath96 if the following condition is satisfied : * standard : for every @xmath112 there exists a @xmath113 such that for all @xmath114 and for all @xmath115 , if @xmath116 then @xmath117 . * non - standard : for all @xmath118 , if @xmath93 then @xmath119 . see remark [ 51b ] for a more detailed discussion . a hyperreal number @xmath4 equal to its own integer part @xmath120\ ] ] is called a hyperinteger ( here the integer part function is the natural extension of the real one ) .
the elements of the complement @xmath121 are called infinite hyperintegers .
let @xmath4 be an infinite hyperinteger .
the interval @xmath2 $ ] has a natural hyperreal extension .
consider its partition into @xmath4 subintervals of equal length @xmath122 , with partition points @xmath123 as @xmath124 runs from @xmath104 to @xmath4 .
note that in the standard setting , with @xmath16 in place of @xmath4 , a point with the maximal value of @xmath38 can always be chosen among the @xmath125 partition points @xmath126 , by induction .
hence , by the transfer principle , there is a hyperinteger @xmath127 such that @xmath128 and @xmath129 consider the real point @xmath130 an arbitrary real point @xmath43 lies in a suitable sub - interval of the partition , namely @xmath131 $ ] , so that @xmath132 . applying `` st '' to the inequality , we obtain by continuity of @xmath38 that @xmath133 , for all real @xmath43 , proving @xmath68 to be a maximum of @xmath38 ( see @xcite ) . we have @xmath134 if and only if whenever the difference @xmath135 is infinitesimal , the difference @xmath136 is infinitesimal , as well , or in formulas : if @xmath137 then @xmath138 . given a sequence of real numbers @xmath139 , if @xmath140 we say @xmath141 is the limit of the sequence and write @xmath142 if the following condition is satisfied : @xmath143 ( here the extension principle is used to define @xmath144 for every infinite value of the index ) .
this definition has no quantifier alternations .
the standard @xmath92-definition of limit , on the other hand , does have quantifier alternations : @xmath145 given a real decimal @xmath146 consider the sequence @xmath147 , @xmath148 , @xmath149 , etc .
then by definition , @xmath150 for every infinite @xmath4 .
now if @xmath151 is a non - terminating decimal , then one obtains a strict inequality @xmath152 by transfer from @xmath153 .
in particular , @xmath154 where the hat @xmath155 indicates the @xmath4-th lightstone decimal place . the standard interpretation of the symbol @xmath0 as @xmath1 is necessitated by notational uniformity : the symbol @xmath156 in every case corresponds to the limit of the sequence of terminating decimals @xmath157 .
alternatively , the ellipsis in @xmath0 could be interpreted as alluding to an infinity of nonzero digits specified by a choice of an infinite hyperinteger @xmath158 .
the resulting @xmath4-infinite extended decimal string of @xmath3s corresponds to an infinitesimally diminished hyperreal value .
such an interpretation is perhaps more in line with the naive initial intuition persistently reported by teachers .
prior to robinson , mathematicians thought of infinitesimals in terms of naive befogging " and vague mystical ideas " .
thus , richard courant @xcite wrote as follows : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ we must , however , guard ourselves against thinking of @xmath159 as an infinitely small quantity " or
infinitesimal " , or of the integral as the sum of an infinite number of infinitesimally small quantities " .
such a conception would be devoid of any clear meaning ; it is only a naive befogging of what we have previously carried out with precision . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ and again on page 101 : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ we have no right to suppose that first @xmath160 goes through something like a limiting process and reaches a value which is infinitesimally small but still not 0 , so that @xmath160 and @xmath161 are replaced by infinitely small quantities " or infinitesimals " @xmath159 and @xmath162 , and that the quotient of these quantities is then formed .
such a conception of the derivative is incompatible with the clarity of ideas demanded in mathematics ; in fact , it is entirely meaningless . for a great many simple - minded people it undoubtedly has a certain charm , the charm of mystery which is always associated with the word infinite " ; and in the early days of the differential calculus even leibnitz [ sic ] himself was capable of combining these vague mystical ideas with a thoroughly clear understanding of the limiting process .
it is true that this fog which hung round the foundations of the new science did not prevent leibnitz [ sic ] or his great successors from finding the right path .
but this does not release us from the * duty of avoiding every such hazy idea * [ emphasis added mk ] in our building - up of the differential and integral calculus .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ needless to say , courant s criticism was not without merit _ at the time of its writing _ ( a quarter century prior to robinson s discovery ) .
i. lakatos @xcite wrote in 66 as follows : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ robinson s work ... offers a rational reconstruction of the discredited infinitesimal theory which satisfies modern requirements of rigour and which is no weaker than weierstrass s theory .
this reconstruction makes infinitesimal theory an almost respectable ancestor of a fully fledged , powerful modern theory , lifts it from the status of pre - scientific gibberish , and renews interest in its partly forgotten , partly falsified history . _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ not everyone was persuaded .
a decade later , courant s _ duty of avoiding every such hazy idea _ was taken up under a constructivist banner by e. bishop . in his essay
@xcite cast in the form of an imaginary dialog between brouwer and hilbert , e. bishop anchors his foundational stance in a species of mathematical constructivism .
thus , bishop s opposition to robinson s infinitesimals , expressed in a bristling review @xcite of keisler s textbook @xcite , was to be expected ( and in fact was anticipated by editor halmos ) . in a memorable parenthetical remark ,
bishop writes @xcite : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ although it seems to be futile , i always tell my calculus students that mathematics is not esoteric : it is common sense .
( even the notorious @xmath163-definition of limit is * common sense * [ emphasis added mk ] , and moreover it is central to the important practical problems of approximation and estimation . ) _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ bishop is referring , of course , to the type of _ common - sense _ definition reproduced in , which he favors over keisler s definition in terms of robinson s hyperreals .
bishop expressed his views about non - standard analysis and its use in teaching in a brief paragraph toward the end of his essay crisis in contemporary mathematics " @xcite . after discussing hilbert s formalist program
he writes : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ a more recent attempt at mathematics by formal finesse is non - standard analysis .
i gather that it has met with some degree of success , whether at the expense of giving significantly less meaningful proofs i do not know .
my interest in non - standard analysis is that attempts are being made to introduce it into calculus courses .
it is difficult to believe that * debasement of meaning * [ emphasis added mk ] could be carried so far . _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ bishop s view of the introduction of non - standard analysis in the classroom as no less than a debasement of meaning " , was duly noted by j. dauben @xcite . to illlustrate how bishop anchors his foundational stance in a species of mathematical constructivism , note that in @xcite
, he writes : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ to my mind , it is a major defect of our profession that we refuse to distinguish [ ... ] between integers that are computable and those that are not [ ... ] the distinction between computable and non - computable , or constructive and non - constructive is the source of the most famous disputes in the philosophy of mathematics ... _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ on page 511 , bishop defines a principle ( lpo ) as the statement that _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ it is possible to search `` a sequence of integers to see whether they all vanish '' , _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ and goes on to characterize the dependence on the lpo as a procedure both brouwer and bishop himself reject .
s. feferman @xcite explains the principle as follows : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ bishop criticized both non - constructive classical mathematics and intuitionism .
he called non - constructive mathematics a scandal " , particularly because of its deficiency in numerical meaning " .
what he simply meant was that if you say something exists you ought to be able to produce it , and if you say there is a function which does something on the natural numbers then you ought to be able to produce a machine which calculates it out at each number .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ the constructivist objections to lpo are similar to objections to the law of excluded middle , and to proofs by contradiction ( for example , the traditional argument for the irrationality of @xmath164 is a proof by contradiction , but using classical continued fraction estimates , it is easy to rewrite it in a numerically meaningful fashion , acceptable from the constructivist viewpoint ) . given that a typical construction of robinson s infinitesimals ( see keisler @xcite ) certainly does rely on lpo , bishop s opposition to such infinitesimals , expressed in a bristling review @xcite of keisler s textbook , was to be expected .
non - standard calculus in the classroom was analyzed in the chicago study by k. sullivan @xcite .
sullivan showed that students following the non - standard calculus course were better able to interpret the sense of the mathematical formalism of calculus than a control group following a standard syllabus .
such a conclusion was also noted by m. artigue @xcite .
in the matter of teaching decimal notation , we would like to obtain some reaction from educators to the following proposal concerning the problem of the unital evaluation of @xmath0 .
a student may ask : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ what does the teacher mean to happen exactly after _ nine , nine , nine _ when he writes _
dot , dot , dot _ ? _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ how is a teacher to handle such a question ?
experience shows that toeing the standard line on the unital evaluation of @xmath165 possesses a high - frustration factor in the classroom . rather than baffling the student with such a categorical claim ,
a teacher could proceed by presenting the following ten points , based on the material outlined in section [ ten ] : 1 .
the reals are not , as the rationals are not , the maximal number system ; 2 . there exist larger number systems , containing infinitesimals ; 3 .
in such larger systems , the interval @xmath2 $ ] contains many numbers infinitely close to @xmath1 ; 4 . in a particular larger system
called the hyperreal numbers , there is a generalized notion of decimal expansion for such numbers , starting in each case with an unbounded number of digits `` @xmath3 '' ; 5 .
all such numbers therefore have an arguable claim to the notation `` @xmath0 '' which is patently ambiguous ( the meaning of the ellipsis `` @xmath166 '' requires disambiguation ) ; 6 .
all but _ one _ of them are strictly smaller than @xmath1 ; 7 .
the _ convention _ adopted by most professional mathematicians is to interpret the symbol `` @xmath0 '' as referring to the _ largest _ such number , namely @xmath1 itself ; 8 .
thus , the students intuition that @xmath0 falls just short of @xmath1 can be justified in a mathematically rigorous fashion ; 9 .
the said extended number system is mostly relevant in infinitesimal calculus ( also known as differential and integral calculus ) ; 10 .
if you would like to learn more about the hyperreals , go to your teacher so he can give you further references .
a goal of our , admittedly non - standard , analysis is both to educate and to heal .
the latter part involves placing balm upon the bewilderment of myriad students of decimal notation , frustrated by the reluctance of their education professionals to yield as much as an infinitesimal iota in their evaluation of the symbol @xmath167 , or to acknowledge the ambiguity of an ellipsis .
we are grateful to david ebin and david tall for a careful reading of an earlier version of the manuscript , and for making numerous helpful suggestions .
abbott , edwin a. : the annotated flatland . a romance of many dimensions .
with an introduction and notes by ian stewart .
reprint of the 2002 original [ perseus publ .
, cambridge ] .
basic books , new york , 2008 .
bishop , e. : review : h. jerome keisler , elementary calculus , _ bull .
* 83 * ( 77 ) , 205 - 208 .
cornu , b. : limits , pp . 153 - 166 , in advanced mathematical thinking . edited by david tall .
mathematics education library , 11 .
kluwer academic publishers group , dordrecht , 1991 .
courant , r. : differential and integral calculus .
i. translated from the german by e. j. mcshane .
reprint of the second edition ( 1937 ) .
wiley classics library . a wiley - interscience publication .
john wiley & sons , inc .
, new york , 1988 .
dauben , j. : appendix ( 1992 ) : revolutions revisited .
7282 in revolutions in mathematics . edited and with an introduction by donald gillies .
oxford science publications . the clarendon press , oxford university press , new york , 1992 .
lakatos , imre : cauchy and the continuum : the significance of nonstandard analysis for the history and philosophy of mathematics .
intelligencer 1 ( 1978 ) , no .
3 , 151161 ( originally published in 66 ) . robinson , abraham : non - standard analysis .
reprint of the second ( 1974 ) edition . with a foreword by wilhelmus a. j. luxemburg .
princeton landmarks in mathematics .
princeton university press , princeton , nj , 1996 .
stewart , i. : from here to infinity .
a retitled and revised edition of the problems of mathematics [ oxford univ .
press , new york , 1992 ] . with a foreword by james joseph sylvester .
the clarendon press , oxford university press , new york , 1996 .
tall , d. : the psychology of advanced mathematical thinking , in advanced mathematical thinking . edited by david tall .
mathematics education library , 11 .
kluwer academic publishers group , dordrecht , 1991 . | is @xmath0 equal to @xmath1 ?
a. lightstone s decimal expansions yield an infinity of numbers in @xmath2 $ ] whose expansion starts with an unbounded number of repeated digits `` @xmath3 '' .
we present some non - standard thoughts on the ambiguity of the ellipsis , modeling the cognitive concept of _ generic limit _ of b. cornu and d. tall .
a choice of a non - standard hyperinteger @xmath4 specifies an @xmath4-infinite extended decimal string of @xmath3s , corresponding to an infinitesimally diminished hyperreal value . in our model , the student resistance to the unital evaluation of @xmath0 is directed against an unspoken and unacknowledged application of the standard part function , namely the stripping away of a ghost of an infinitesimal , to echo george berkeley .
so long as the number system has not been specified , the students hunch that @xmath0 can fall infinitesimally short of @xmath1 , can be justified in a mathematically rigorous fashion . supported by the israel science foundation ( grants no . 84/03 and 1294/06 ) and the bsf ( grant 2006393 ) ] |
graphs appear in many current applications . in social sciences
groups of people are often modeled by letting the vertices in the graph represent persons and edges represent the interactions or relationships between them .
edges can be directed or undirected , the later indicating a reciprocal relationship between the vertices .
usually the graphs created from such datasets are simplifications of the original dataset .
one typical simplification is to allow only directed or only undirected edges .
however , in real world graphs it is common to find a combination of directed and undirected edges . in @xcite we find some examples of empirical graphs where the proportion of directed edges is in the range 0.26 - 0.85 , the rest being undirected edges .
additional examples are shown in _ table [ tab : directed ] _ where the proportion of directed edges has been calculated for some social networks that can be found in @xcite .
we expect such graphs to be better represented by _ partially directed graphs _
, where we allow both directed and undirected edges .
.proportion of directed edges for some data sets from @xcite , when viewed as partially directed graphs .
we see that several of these graphs have a substantial proportion of undirected edges and of directed edges , such that neither type should be ignored .
[ cols="<,>,>,>",options="header " , ] the _ configuration model _ has been used extensively to model undirected networks @xcite .
it has also been been adapted to work for directed graphs @xcite . in the configuration model
the graph is constructed by first assigning a degree to each vertex of the graph and then connecting the edges uniformly at random .
the degrees of the vertices of the graph are either given as a degree sequence or the degrees are drawn from some given degree distribution .
graphs created in this way will share some properties with real world graphs , but will be different in other aspects .
e.g. the configuration model for directed networks will have a very low proportion of reciprocal edges , i.e. two parallel directed edges in opposite directions .
this is an effect of connecting edges uniformly at random in this type of graph .
this can be undesirable if we wish to use the configuration model graph as a _ null reference _ to compare with a real - world graph .
while we wish to connect the edges uniformly at random , we may want to preserve the degree distribution , including any dependence between the indegrees , outdegrees and undirected degrees .
in this paper we consider a partially directed configuration model where we allow both directed and undirected edges .
any vertex in such a partially directed configuration model graph can have all three types of edges : _ incoming _ , _ outgoing _ and _ undirected_. we select the degree of each vertex from a given joint , three dimensional degree distribution and we do not assume or require the in- , out- and undirected degrees to be independent .
when connecting the edges , outgoing edges can only connect to incoming edges and undirected edges can only connect to undirected edges . once all edges are connected we make the graph _ simple _ and thus do not allow self loops or parallel edges of any type .
we make the graph simple by erasing conflicting edges and by converting parallel undirected edges in opposite directions into undirected edges .
since this process modifies the degree of some of the vertices , it is not certain that the empirical degree distribution converges to the degree distribution we started with .
however , in _ section [ sec : mainresults ] _ we show that , with suitable restrictions on the first moments of the degree distribution , the degree distribution asymptotically converges to the desired one .
note that , by selecting a joint degree distribution in the proper way we can also create completely directed graphs or completely undirected graphs , with or without any dependence between the degrees .
thus the presented partially directed configuration model incorporates several of the already existing models . in the next section , _ section [ sec : mainresults ] _ , we present definitions and state the main result of the paper . detailed derivations and proofs have been postponed to _ section [ sec : proofs]_. to illustrate how these graphs work , _ section [ sec : examples ] _ is devoted to some simulations of partially directed graphs , showing results for small and for large @xmath0 .
the latter is to give an intuitive feeling for the asymptotic results and the former is to illustrate that significant deviations from these asymptotic results are possible for small @xmath0 .
a comparison with an empirical social network is also done .
conclusions and discussion can be found in _
section [ sec : concl]_.
in this section we define the configuration model for partially directed graphs .
we define the terminology used , how the graph is created from a degree distribution , how the graph is made simple and finally show , with suitable restrictions on the first moments of the degree distribution , that the degree distribution of the partially directed configuration model graph asymptotically converges to the desired distribution .
proofs are left for _ section [ sec : proofs ] _ a graph consists of vertices and of edges .
the size of the graph , the number of vertices , is denoted @xmath0 . here
we will specifically study the case when @xmath1 .
we work with graphs that are _ partially directed _ , meaning that any vertex can have incoming edges , outgoing edges and undirected edges .
we distinguish between edges and _
stubs_. by stubs we mean yet unconnected half - edges of a vertex . in the same way as edges
, stubs can be in - stubs , out - stubs and undirected stubs .
the number of stubs of the different types is the degree of a vertex and will be denoted @xmath2 , where the individual terms represent the indegree , outdegree and undirected degree , respectively .
when the degree of the vertex is a random quantity , it is denoted @xmath3 .
a degree sequence that is non random is denoted @xmath4 , @xmath5 , where @xmath0 is the number of vertices in the graph . when these degree sequences are random vectors they are denoted @xmath6 .
degrees can be assigned to the vertices from some given joint degree distribution with distribution function @xmath7 for which the probability of a specific combination of indegree , outdegree and undirected degree is called @xmath8 .
we will also use the marginal distributions .
we have @xmath9 for the incoming edges , @xmath10 for the outgoing edges and @xmath11 for the undirected edges .
the corresponding random variables , i.e. the number of edges of each type , will be denoted @xmath12 , @xmath13 and @xmath14 .
other quantities of interest are the moments of the distribution . here
we will consider the first moments @xmath15=\sum i{p^{\leftarrow}}_{i } $ ] , @xmath16=\sum j{p^{\rightarrow}}_{j } $ ] and @xmath17=\sum k{p^{\leftrightarrow}}_{k } $ ] .
a graph is _ simple _ if there are no unconnected stubs , no self - loops and no parallel edges . for a finite graph of size
@xmath0 we also want to count the number of vertices with a certain degree @xmath18 .
we call this quantity @xmath19 . dividing by @xmath0 we can calculate @xmath20 , the proportion of the number of edges that have degree @xmath18 . whenever the graph is created by some random process , we can also consider the expectation of this random quantity @xmath21 $ ] , which defines the distribution function @xmath22 . we define the partially directed configuration model as follows : 1 .
we start with a graph with @xmath0 vertices , but without any edges or stubs .
2 . for each vertex
, we independently draw a degree @xmath23 from @xmath7 at random .
we connect undirected stubs with other undirected stubs .
we do this by picking two undirected stubs uniformly at random and connecting them .
we repeat this with the remaining unconnected undirected stubs until there is at most one undirected stub left .
we connect directed incoming stubs with directed outgoing stubs .
we do this by picking one directed incoming stub and one directed outgoing stub , both independently and uniformly at random and then connecting them .
we repeat this with the remaining unconnected directed stubs until we are out of incoming stubs or outgoing stubs ( or both ) .
we want the graph to be simple , but the connection process may have left some stubs unconnected and may also have created self - loops and parallel edges .
we make the graph simple by erasing some stubs and edges .
we define the procedure in such a way that the connectivity of the graph is maintained : 1 .
erase all unconnected stubs .
there can be at most one unconnected undirected stub , while there may be a larger number of unconnected directed stubs , either all incoming or all outgoing , if the number of in - stubs is not equal to the number of out - stubs .
2 . erase all self - loops , both directed and undirected .
3 . when there are parallel _ identical _ edges , erase all except one of them .
4 . erase all directed edges that are parallel to an undirected edge .
erase each pair of reciprocal directed edges and add a single undirected edge instead .
while this step decreases the number of directed edges , it also increases the number of undirected edges . from the above description
we see that there are two non - deterministic steps that affect the degrees of the vertices in the creation of the simple partially directed graph : 1 . assigning degrees from the distribution @xmath7 .
2 . connecting the stubs uniformly at random . while this does not , in itself , modify the degrees of the vertices , it affects which stubs and edges that will be erased when making the graph simple .
this process results in a finite graph for which the value of @xmath24 had been closer to 2 , then the average number of deleted stubs would not have decreased as clearly as it does now , indicating that the average number of deleted edges then decreases only slowly with the graph size .
this still would not in itself contradict convergence in distribution since a large proportion of the deleted edges can then be contributed to a small number of vertices of high degree , and so would not affect the overall convergence of the degree distribution.ch the degree distribution can not be expected to be the same as @xmath7 .
however , we later show that , with suitable restrictions on the distribution @xmath7 , the distribution @xmath22 that was defined above , asymptotically approaches @xmath7 .
the results in this section are inspired by , and to some degree follow @xcite .
the theorem establishes the asymptotic convergence of the degree distribution .
[ thm : distribution ] if @xmath7 has finite mean for each component , so @xmath25 , @xmath26 , and @xmath27 , and also @xmath28 then , as @xmath1 1 .
@xmath29 , 2 .
@xmath30 , that is , the empirical distribution converges in probability to f. the proof , which is postponed to _ section [ sec : proofs ] _ , follows the same line of reasoning as in @xcite , but with modifications to take into account the complications introduced by allowing both directed and undirected edges in the graph .
although _ theorem [ thm : distribution ] _ establishes the asymptotic convergence of the degree distribution , it remains to see how how well this holds for finite graphs . in this section
we investigate this by looking at a scale - free distribution , at a poisson degree distribution and at an empirical network . since we are working with a joint degree distribution , in addition to the distribution for each of the three stub types we also need to consider the possible dependence between the different types .
_ table [ tab : examples ] _ gives an overview of how the data for the plots were created .
@p0.09p0.48p0.16p0.16@ degreedistribution & method & independent & dependent + empirical & empirical degree data from the dataset @xcite was used .
this is data from an on - line social site .
some characteristics can be found in _ table [ tab : directed]_. the mean degrees for in - degrees , out - degrees and undirected degrees are 3.6 , 3.6 and 10.6 , respectively ( not shown ) . when viewed as a directed graph and counting all stubs this gives a total mean degree of approximately 28.3 . here
we count the undirected edges as two edges since it consists of an incoming and an outgoing edge , when viewed as an edge in a directed graph .
both the directed and the undirected edges have degree distributions that are _ approximately _ scale - free in the tail , with @xmath31 and @xmath32 ( not shown ) . & each stub type is treated individually and independent samples are drawn , with replacement , for each vertex and each stub type . &
independent samples of complete vertices are drawn , with replacement , from the pool of empirical vertices .
+ scale - free & the selected distribution function is @xmath33 with @xmath34 where @xmath35 is the riemann zeta function .
the tail of this distribution is asymptotically @xmath36 .
this specific distribution function was selected because of its scale - free property , while still being easy to simulate from using a discrete variant of the inverse transformation method ( * ? ? ?
* see section 11.2.1 and also example 11.7 ) ) . for all simulations @xmath37 , which is the coefficient for the directed edges in the empirical graph .
this value gives finite expectation ( approximately 2.7 ) , but infinite variance .
this is consistent with the assumptions in _ theorem [ thm : distribution]_. & for each vertex and each stub type an independent sample from the assigned distribution was drawn . & for each vertex an independent sample from the assigned distribution was drawn and the same degree was assigned to all stubs for the vertex .
+ poisson & degrees drawn from poisson distribution with parameter 7 , thus having mean degree 7 .
when treated as a directed graph and counting all stubs the total mean degree is 28 , close to the value 28.3 for the empirical graph above .
& see above . & see above .
+ since _ theorem [ thm : distribution ] _ focuses on showing convergence to the correct degree distribution , studying the total variation distance , @xmath38 ( defined in _ section [ sec : totvardist ] _ ) , is of interest ( see e.g. @xcite ) .
we also study the number of erased edges as a function of the graph size .
finally , we study the size of the strongly connected giant component and the distribution of small components for a few different graphs based on the empirical data from livejournal .
the dataset livejournal1 @xcite is a directed graph created from the declaration of friends in a social internet community .
the original graph contains self loops , but these have been removed in this analysis .
the simple graph has a proportion of directed edges of about 0.4 , so this is a good example of a graph where both directed and undirected edges play an important role .
when sampling from this distribution to create the configuration model graph , the degrees of vertices from the original ( partially directed ) graph were drawn independently and uniformly at random , with replacement .
thus the frequencies of the degrees found in the graph were used as the given distribution @xmath7 and this distribution function is then compared with the distribution @xmath22 created by sampling from @xmath7 , connecting the edges and making the graph simple . _
theorem [ thm : distribution ] _ states that @xmath30 and thus we define the following version of the total variation distance : @xmath39 where the @xmath40 is introduced so that @xmath38 can only take on values in the range @xmath41 $ ] . as @xmath1
we expect to see that the total variation distance tends towards zero . when we generate the graphs according to the configuration model we replace @xmath19 with the corresponding empirical sample @xmath42 from one realization of a random graph .
we can then repeat this process with more samples of random graphs and plot this .
the result is shown in _ figure [ fig : totvardist ] _ , where we have also taken the average of the empirical total variation distance for 100 random graph samples . in _ figure
[ fig : totvardist ] _ we see that the total variation distance decreases towards zero .
the fastest decrease is for the poisson graph , and the reason is that this distribution has a light tail when compared with the scale - free distribution .
a closer look att the empirical graph reveals that the distributions for the directed and the undirected edges look much like a scale - free distribution .
the in- and the out - degree have @xmath43 and the undirected degree has @xmath44 in the tail ( not shown ) .
thus the tail for the empirical distribution is heavier than for the poisson distribution and so we can expect a slower convergence for the empirical graph .
even slower convergence has been observed ( not shown ) for values of @xmath24 even closer to 2 , e.g. try @xmath45 .
this is not surprising as the distribution then becomes more heavy - tailed . if we continue even further , to @xmath46 the conditions used in the proof of _ theorem [ thm : distribution ] _ no longer hold , since the expectations are no longer finite , and thus we should not expect the total variation distance to converge to zero for these values of @xmath24 . from the figure we also see that the _ dependent _ curve for the poisson distribution is clearly lower than the _ independent _ curve .
one explanation for this is that when the degrees for in - stubs and the out - stubs are identical for each vertex , as in the dependent graph ( as defined in _
table [ tab : examples ] _ ) , the total number of in - stubs will be equal to the total number of out - stubs and thus no directed stubs will be deleted for this reason
. there may still be self - loops and parallel edges , but for the poisson graph these are few compared to the number of stubs deleted in the independent graph ( as defined in _
table [ tab : examples ] _ ) where there is a mismatch between the number of in - stubs and the number of out - stubs .
for the empirical graph and for the scale - free graph the same phenomenon can not be observed .
one explanation to this is that the scale - free independent model is not dominated by the deletion of leftover directed edges .
instead the number of self - loops and parallel edges are of the same order of magnitude as the leftover directed edges ( see _ figure [ fig : reldeledges ] _ ) .
thus the difference between the curves for the total variation distance is much smaller for the scale - free and for the empirical graph .
another answer to why the empirical graph does not show a big difference between the dependent and the independent curve can be that the dependent version of the empirical graph does not have the same type of complete dependence as for the scale - free or the poisson graph . in the empirical dependent graph ,
degrees are assigned by sampling the degrees of vertices from the original empirical graph , and thus the number of in - stubs will in general not equal the number of out - stubs . looking at _ figure
[ fig : reldeledges ] _ we see that the number of directed unconnected edges is almost the same for the independent version as for the dependent version of the empirical graph .
looking instead at the same plot for the poisson graph we note that the deletion of directed unconnected stubs dominates the independent version of the graph , while there are no such deleted stubs in the dependent version of the graph .
0.48 , for the po(7 ) model and for the empirical configuration model .
each data point shows the average of 100 simulations.,title="fig:",scaledwidth=100.0% ] 0.48 , for the po(7 ) model and for the empirical configuration model .
each data point shows the average of 100 simulations.,title="fig:",scaledwidth=100.0% ] 0.48 , for the po(7 ) model and for the empirical configuration model .
each data point shows the average of 100 simulations.,title="fig:",scaledwidth=100.0% ] 0.48 , for the po(7 ) model and for the empirical configuration model .
each data point shows the average of 100 simulations.,title="fig:",scaledwidth=100.0% ] 0.48 , for the po(7 ) model and for the empirical configuration model .
each data point shows the average of 100 simulations.,title="fig:",scaledwidth=100.0% ] 0.48 , for the po(7 ) model and for the empirical configuration model .
each data point shows the average of 100 simulations.,title="fig:",scaledwidth=100.0% ] the number of erased edges will depend on the degree distribution , on the graph size and will also be different each time a graph is created according to the configuration model . in _ figure [ fig : reldeledges ] _ the average number of erased edges per vertex were plotted .
each point corresponds to the average of 100 simulations of random graphs according to the partially directed configuration model .
the deleted edges were classified as to the reason why they were deleted as defined in the rules in _ section [ sec : creatinggraph]_. for all plots , the graphs indicate that the average number of deleted stubs or edges per vertex decreases with the size of the graph .
thus also the risk of any vertex having its degree affected by the deletion of a stub or an edge goes down and this indicates that the degree distribution @xmath22 converges to @xmath7 asymptotically .
the scale - free distribution is more difficult since for @xmath46 neither the variance nor the expectation exist .
here we have selected @xmath37 for the scale - free graph .
this value gives finite expectation , but infinite variance . as already briefly mentioned in _ section [ sec : totvardist ] _
, for the scale - free and for the poisson curves there are no deleted directed stubs for the dependent plots .
this is because of how the dependent graphs are created . in these graphs ,
each vertex has the same number of in - stubs and out - stubs .
thus there will not be any directed stubs left over after the graph has been connected so no such stubs will be deleted . for the empirical graph
this is not the case since the dependent version of the graph is created by sampling from the empirical degrees of the vertices , and for these the number of in - stubs in general do not equal the number of out - stubs .
in fact we note that the average number of deleted directed stubs per vertex seem to be approximately equal for the directed and the undirected version of the empirical graph , possibly indicating a quite poor correlation between in - stubs and out - stubs in the original graph .
another difference between the graphs is that for the scale - free dependent graph there are many more deleted directed reciprocal edges , deleted directed self loops and deleted directed edges that are parallel with undirected edge , compared with the independent scale - free graph .
this can be explained by the heavy tail of the scale - free distribution .
for instance assume that some vertex has a very high degree .
since the degrees are dependent ( equal , in this case ) , the risk is much higher that there will be self loops among the directed edges .
also , since the undirected degree will also be high for this vertex , the risk of having directed edges in parallel with the undirected edges also increases . finally the chance of getting reciprocal directed edges also increases .
this risk is high if there are many vertices with high degrees . in the dependent case
if two vertices have many in - stubs both will also have many out - stubs , increasing the chance of parallel edges between these . )
.,scaledwidth=70.0% ] finally we study the strongly connected components in the original data from livejournal , compared with the configuration model based on partially directed stubs and also on directed stubs .
for any vertex @xmath47 we define the out - component of vertex @xmath47 as the set of all vertices that can be reached from vertex @xmath47 by following the edges of the graph and respecting how they are directed . in the same way
the in - component of vertex @xmath47 is defined as the set of all vertices from which we can reach vertex @xmath47 .
the intersection of the out - component and the in - component defines the strongly connected component of vertex @xmath47 .
any two vertices @xmath48 where we can reach @xmath49 from @xmath47 and @xmath47 from @xmath49 have the same strongly connected component .
thus the graph can be uniquely divided into a set of strongly connected components .
here we study the strongly connected components of the empirical graph and also of configuration model graphs created by using the degree sequence of the empirical graph as the given degree distribution .
the largest component in the graph corresponds to the notion of a giant component , the size of which is proportional to the size of the graph .
the size of the giant component for these simulations can be compared with theoretical results for a configuration model graph with given degree distribution ( see @xcite ) . by plugging in the empirical degree distribution of the livejournal dataset
, we get that the theoretical size of the giant component 0.8040 for the partially directed graph , and 0.8028 for the directed graph .
these values show a good match with the simulation data presented in _ figure [ fig : components]_. it is not surprising that the largest component is largest in the configuration model for the partially directed graph .
the original empirical graph is likely to have sub - communities that may connect only weakly to other communities , thus reducing the total size of the largest strongly connected component , but of course increasing the number of moderately sized strongly connected components .
the directed graph lacks the undirected edges and thus the largest strongly connected component will not include vertices that are connected to it only via a directed edge ( in one direction only ) .
thus its largest strongly connected component will be smaller than for the partially directed graph .
when looking at the variation in size among the medium sized components in _ figure [ fig : components ] _ , this is largest for the original empirical graph . for the configuration model on the directed graph _ all _ other components consist only of single vertices , while for the configuration model on the partially directed graph components of size 1 - 4 exist .
the appearance of some larger small components for the partially directed graph is caused by the undirected edges , compared with only directed edges for the completely directed graph , as was already mentioned above .
in this section we provide a proof of _ theorem [ thm : distribution]_. the first part of the proof closely follows @xcite , with modifications for the joint distribution . in @xcite
the proof is for the undirected graph , and the addition of the directed edges makes things more complicated .
there are mainly two things that need more detailed treatment , the 3-dimensional degree distribution and the fact that combining undirected and directed edges in the same graph creates new reasons for why edges are erased , affecting the empirical degree distribution and thus also , possibly , the asymptotic behavior of it . the first part of the proof , that is similar to @xcite has been moved to two lemmas ( _ lemma [ lem : b - a ] _ and _ lemma [ lem : sufficient ] _ ) to make the part of the proof that is specific for the partially directed configuration model graph more accessible .
a third lemma ( _ lemma [ lem : convergence ] _ ) that helps in the final part of the proof of _ theorem [ thm : distribution ] _ has also been included .
[ lem : b - a ] @xmath30 implies @xmath29 as @xmath1 .
a. @xmath50 and @xmath51 imply @xmath52\to p_{{\underline{d } } } $ ] , by bounded convergence @xcite .
b. @xmath52=p_{{\underline{d}}}^{(n ) } $ ] implies @xmath53 c. since ( ii ) is valid for any @xmath18 we have @xmath29 as @xmath1 . in _ lemma [ lem : sufficient ]
_ we need a few definitions that are used both in the lemma and in the proof of it .
let @xmath54 be an indicator variable that shows if vertex @xmath55 has had its degree modified in the process of creating a simple configuration model graph of size @xmath0 .
the total number of modified vertices can then be calculated by summing all of these and we define @xmath56 . [
lem : sufficient ] if @xmath57 and for arbitrary @xmath55 , then @xmath50 as @xmath1 .
a. let @xmath58 be the number of vertices with degree @xmath18 before any stub has been erased or added . by the law of large numbers we have that @xmath59 as @xmath1 .
since we want to show that @xmath50 it is enough to show that @xmath60 as @xmath1 .
b. we note that modifying the degree of a vertex affects not only the number of vertices with the original degree , but also the number of vertices with the new degree , thus @xmath58 can be less than @xmath19 .
however , we can still be sure that @xmath61 .
we wish to show that @xmath62 , i.e. that @xmath63 as @xmath64 .
c. using markov s inequality and that @xmath65 we get @xmath66}{\epsilon};\forall\epsilon>0.\ ] ] thus it is enough to show that @xmath67\to 0 $ ] .
d. the @xmath68 are identically distributed since the numbering of the vertices is arbitrary and so @xmath69={\mathrm{e}}\left[{m_{1}^{(n)}}\right]={\mathrm{p}}\left({m_{1}^{(n)}}{=}1\right ) $ ] , where vertex 1 has been chosen arbitrarily .
we want to show that @xmath70 or , equivalently , that @xmath71 as @xmath1 .
e. conditioning on the degree of vertex 1 gives @xmath72 since we know @xmath73 it is enough to show that @xmath74 [ lem : convergence ] let @xmath75 be a sequence of non - negative random variables and let @xmath76 be a non - negative random variable .
also let @xmath77 be a real number .
if @xmath78 , @xmath79\leq a $ ] and @xmath80=a $ ] , then @xmath79=a $ ] . for non - negative random variables @xmath81 fatou s lemma states @xmath82\leq \liminf_{m\to\infty}{\mathrm{e}}\left[y_m\right].\ ] ]
we apply skorokhod s representation theorem and can thus define @xmath81 and @xmath83 ( all on the same probability space ) to have the same distribution as @xmath75 and @xmath76 , and @xmath84 as @xmath85 developing the left and right hand side of fatou s lemma now gives : @xmath86 = { \mathrm{e}}[y]={\mathrm{e}}[x]=a,\ ] ] @xmath87\leq\lim_{n\to\infty}{\mathrm{e}}[y_m]=\lim_{m\to\infty}{\mathrm{e}}[x_m]\leq a.\ ] ] thus @xmath88=a\ ] ] now we are ready to prove the main theorem . 1 .
_ lemma [ lem : b - a ] _ shows that _ theorem [ thm : distribution ] _ _ ( b ) _ implies _
( a)_. 2 .
it remains to prove _ theorem [ thm : distribution ] _ _ ( b)_. _ lemma [ lem : sufficient ] _ simplifies this process .
+ let @xmath89 be the _ indicator _ variable for the event that a specific vertex ( arbitrarily selected to be vertex 1 ) has had its degree _ modified _ when creating a simple configuration model graph of size @xmath0 according to the procedure defined in _ section [ sec : creatinggraph]_. also let the degree of vertex 1 be @xmath90 . according to _ lemma [ lem : sufficient ] _ , in order to prove ( b ) it is sufficient to show that @xmath91 3 . remembering that we do not allow self loops or parallel edges , @xmath92 exactly when each stub from vertex 1 is saved . in total , vertex 1 has @xmath93 stubs and these are all saved only when all of them successfully attach to other matching stubs , all from different vertices selected from vertices @xmath94 . in all other cases
the degree of vertex 1 will surely be modified , giving no contribution to the probability of @xmath95 .
+ now , if we knew the degrees of all the vertices , it would be easy to calculate the probability of @xmath95 .
we do this simply by considering _ all _ events where the stubs of vertex 1 connect to different vertices and then sum all the probabilities of these events .
it is thus natural to continue the proof by conditioning on the degrees of vertices @xmath94 .
let the degrees of vertices @xmath94 be @xmath96 , where the @xmath23 are i.i.d . from @xmath7 .
then we want to study @xmath97 .
\label{eqn : saveprob}\ ] ] 4 .
we now look more closely at the conditional probability @xmath98 + where @xmath99 is a specific outcome of the degrees of the vertices . from this
we see that the total number of stubs of each type are @xmath100 , @xmath101 and @xmath102 .
we want to know where each stub of vertex 1 _ attempts _ to connect and define a set set of indices , @xmath103 , @xmath104 and @xmath105 .
any set of values of these indices we call a _ save - attempt _ , indicating that we try to save all stubs of vertex 1 from being erased , by attempting to connect the stubs of vertex 1 to matching stubs from the vertices pointed to by these indices . + given the degrees of all vertices we can calculate the probability of any such _ save - attempt_. first some basic observations : 1 . if any one of the selected vertices do not have a matching stub the probability of the _ save - attempt _ is zero . as an example , assume that an in - stub attempts to connect to vertex 2 , but vertex 2 does not have any out - stub at all
. then this event will have probability zero .
2 . as a consequence , for the _ save - attempt _ to have a probability larger than zero , _ all _ the vertices that the stubs of vertex 1 attempt to connect to must have matching stubs . + as an example , take a look at the _ save - attempt _ where each stub of vertex 1 tries to connect to the other vertices in order .
the indices then take on the values @xmath106 . for now , we ignore the probability that there may not be enough matching stubs of vertices @xmath107 to accommodate all the stubs of vertex 1 .
we do this now to make the main argument clearer , but we correct the equations for this special case later in the proof .
+ first we look at in - stub 1 from vertex 1 .
since we are working with the configuration model , this stub has an equal chance of connecting to any of the matching stubs .
thus the probability of in - stub 1 from vertex 1 to connect to any of the out - stubs from vertex 2 is @xmath108 + once in - stub 1 of vertex 1 has connected to vertex 2 we continue with in - stub 2 of vertex 1 . once again the configuration model tells us that this stub has an equal chance of connecting to any of the remaining matching stubs .
thus the probability of it connecting to any of the out - stubs from vertex 3 is @xmath109 + we can continue in the same way with the rest of the in - stubs , then the out - stubs and finally the undirected stubs of vertex 1 . for the undirected stubs we note that we need to subtract 2 stubs every time we connect one stub , since the undirected stubs connect to other undirected stubs .
+ now we can calculate the probability of this specific _ save - attempt _ and find that it is @xmath110 + in the expression we have ignored that we we have already used up @xmath111 out - stubs when connecting the in - stubs of vertex 1 .
we correct for this in the final expressions given later in the proof .
+ here we explicitly see that this expression is equal to zero iff any one of the degrees in the numerator is zero
. otherwise it will be positive , but always less than or equal to 1 .
+ to shorten the expressions we will call each of the three parts of _ eq .
( [ eqn : specprob ] ) _
@xmath112 , @xmath113 and @xmath114 , respectively , where the arrow indicates what type of stub in vertex 1 we are dealing with .
+ now we are ready to write down the expression for the conditional probability in _ eq .
( [ eqn : condprob ] ) _ we need to sum _ eq .
( [ eqn : specprob ] ) _ over _ all _ values of @xmath115 , @xmath116 and @xmath117 , such that _ all _ sub - indices are different - pointing to different vertices .
we arrive at @xmath118 + the number of terms in the sum will be @xmath119 , which is simply the number of different ways in which we can select the @xmath120 indices out of the @xmath121 possible vertices . note that these combinations of indices include the ones we are interested in , where all stubs of vertex 1 are saved .
note also that the sum includes some combinations that we are not interested in , but all of these have probability zero and so it does not matter if we include them in the sum or not . 5 .
we now need to deal with a few complications that will lead to corrections to @xmath112 , @xmath113 and @xmath114
if the number of stubs of vertex 1 ( @xmath120 ) is larger than the number of available nodes ( @xmath121 ) , then it is not possible to select all sub - indices indices different .
however , since @xmath120 is fixed , this is always resolved as @xmath1 . in the following
we will always assume that @xmath122 .
2 . there may be a mismatch in the number of stubs .
if the number of undirected stubs is odd , there will be one extra stub .
let @xmath123 be the number of such stubs .
clearly @xmath123 can only be 0 or 1 .
+ in the same way the number of in - stubs may differ from the number of out - stubs .
let @xmath124 , the difference between the number of in - stubs and the number of out - stubs .
clearly @xmath125 can be negative , zero or positive .
if @xmath123 or @xmath125 are not zero then some stubs will remain unconnected .
+ in the following we will deal with both of these by imagining two extra _ pools _ of edges each of size @xmath123 and @xmath126 , respectively .
these pools behave just as any normal vertex and any stub has an equal probability to connect to any allowed stub , including these two pools .
they are thus added to the denominators in _ eq .
( [ eqn : specprob ] ) _ 3 .
as mentioned before , we have included some events that have probability zero in the sum .
although the nominator is always zero for these , in some special cases the denominator may also become zero .
this happens when there are not enough matching stubs to accommodate all the stubs of vertex 1 .
of course we could define @xmath127 , but here we instead chose to correct the denominator so that it does not become zero .
we do this correction by adding an extra indicator variable to the denominator . whenever this happens , the nominator is still zero , so the sum is not changed .
+ the corrected versions of @xmath112 , @xmath113 and @xmath114 are thus @xmath128 6 . to be able to obtain an expression for the probability in _ eq .
( [ eqn : condprob ] ) _ we need to replace the degrees in _ eq .
( [ eqn : corrspecprob ] ) _ with their stochastic counterpart to obtain @xmath129 where @xmath130 here the uppercase variables are all the stochastic counterparts of the lowercase variables defined previously . 7 .
now we can continue with _ eq .
( [ eqn : saveprob ] ) _ : @xmath131 \\ & = { \mathrm{e}}\left [ \sum_{\substack{{\mathbf{i}},\ , { \mathbf{j}},\ , { \mathbf{k}}\\ \text{all sub - indices different}}}{{q^{\overset{(n)}{\leftarrow}}_{{\mathbf{i}}}}{q^{\overset{(n)}{\rightarrow}}_{{\mathbf{j}}}}{q^{\overset{(n)}{\leftrightarrow}}_{{\mathbf{k } } } } } \right ] \\ & = \sum_{\substack{{\mathbf{i}},\ , { \mathbf{j}},\ , { \mathbf{k}}\\ \text{all sub - indices different}}}{{\mathrm{e}}\left [ { q^{\overset{(n)}{\leftarrow}}_{{\mathbf{i}}}}{q^{\overset{(n)}{\rightarrow}}_{{\mathbf{j}}}}{q^{\overset{(n)}{\leftrightarrow}}_{{\mathbf{k}}}}\right ] } \\ & = ( n-1)(n-2)\cdot ...
\cdot(n - d){\mathrm{e}}\left [ { q^{\overset{(n)}{\leftarrow}}_{{\mathbf{i}}}}{q^{\overset{(n)}{\rightarrow}}_{{\mathbf{j}}}}{q^{\overset{(n)}{\leftrightarrow}}_{{\mathbf{k}}}}\right ] \\ & = \frac{(n-1)\cdot ...
\cdot(n - d)}{n^{d}}{\mathrm{e}}\left[n^{d}{q^{\overset{(n)}{\leftarrow}}_{{\mathbf{i}}}}{q^{\overset{(n)}{\rightarrow}}_{{\mathbf{j}}}}{q^{\overset{(n)}{\leftrightarrow}}_{{\mathbf{k}}}}\right ] \\ & = c^{(n ) } { \mathrm{e}}\left [ \left(n^{{d^{\leftarrow}}}{q^{\overset{(n)}{\leftarrow}}_{{\mathbf{i}}}}\right )
\left(n^{{d^{\rightarrow}}}{q^{\overset{(n)}{\rightarrow}}_{{\mathbf{j}}}}\right ) \left(n^{{d^{\leftrightarrow}}}{q^{\overset{(n)}{\leftrightarrow}}_{{\mathbf{k}}}}\right ) \right ] .
\label{eqn : condprobfinal } \end{aligned}\ ] ] + _ note 1 : _ the expectation and the summation can be interchanged since all terms are non - negative and since the summation does not depend on any random quantity ( as mentioned before ) .
+ _ note 2 : _ since vertex degrees are drawn independently at random , all expectation terms in the sum are identical and we simply take the number of terms times the expectation of one of the terms instead of the sum .
the number of terms was already discussed above .
+ _ note 3 : _
all that remains is to take the limit of _ eq .
( [ eqn : condprobfinal])_. we start by studying the limit of what is inside the expectation . rewriting the first term we get @xmath133
+ the remaining outgoing and undirected terms will be very similar , producing the additional terms @xmath134 , @xmath135 , @xmath136 , @xmath137 and @xmath138 in the denominator .
+ now note that , since @xmath139 and @xmath140 , we have that @xmath141}\leq1\ ] ] by the law of large numbers and using slutsky s theorem @xmath142 here we used that @xmath143 , by assumption .
+ since all @xmath144 , @xmath145 and @xmath146 are independent by construction we also have @xmath147 = 1.\ ] ] now we use _ lemma [ lem : convergence ] _ and immediately conclude that @xmath141}=1\ ] ] and thus also that @xmath148 } = 1.\ ] ] this is what we wanted to show and so the proof is complete .
we have shown a simple way to create a partially directed configuration model graph from a given joint degree distribution .
the graph is simple , and under specified conditions the degree distribution converges to the desired one .
the proof is generic and can be extended to any type of graph where stubs are saved from being erased if they connect to other ( unique ) vertices .
the only assumptions in the proof are that the degrees of different vertices are independent , that the expectation of the degree of each type of stub is finite and that the expectation of the degree for the in - stubs is equal to the expectation for the degree of the out - stubs .
this means that the proof works also for undirected graphs and for directed graphs , and also if the number of different types of stubs is increased to any finite number , as long as similar conditions as in this proof are fulfilled .
allowing for self loops and parallel edges only increases the chance of saving a stub from being erased and so is not a problem .
the main advantage of using a partially directed model to represent empirical networks , as opposed to using a completely directed or completely undirected model , is that the partially directed model preserves the proportion of undirected edges .
this is important for networks where there is a significant proportion both of directed and of undirected edges , and where none of the different types of edges can be ignored .
examples of such graphs have been given in _ table [ tab : directed]_. the model also preserves any dependence between directed and undirected degrees present in the original empirical graph or the given degree distribution .
however , this model does not produce other structures that can often be found in empirical networks .
e.g. it does not produce the same number of moderately sized strongly connected components that we see in the empirical networks . in this respect it does however perform slightly better than the configuration model on directed graphs . possible improvements towards realism would be to see how e.g. triangles ( of different types ) , different types of vertices and other heterogeneities could be included in the model .
k.s . was supported by the swedish research council , grant no .
2009 - 5759 .
t.b . is grateful to riksbankens jubileumsfond ( contract p12 - 0705:1 ) for financial support . | the configuration model was originally defined for undirected networks and has recently been extended to directed networks .
many empirical networks are however neither undirected nor completely directed , but instead usually partially directed meaning that certain edges are directed and others are undirected . in the paper
we define a configuration model for such networks where nodes have in- , out- , and undirected degrees that may be dependent . we prove conditions under which the resulting degree distributions converge to the intended degree distributions .
the new model is shown to better approximate several empirical networks compared to undirected and completely directed networks . |
when two ultra - relativistic nuclei collide , they leave behind a region of high energy - density qcd matter , whose properties we would like to understand better .
the initial geometry of the qcd matter is set by the overlap region of the two colliding nuclei .
generally , the nuclei collide at finite impact parameter rather than head - on . in this case
the initial geometry is not a disk , but is an `` almond shaped '' ellipse .
( the short and long axis of the initial almond are taken as the @xmath4 and @xmath5 axes respectively . )
the production mechanism of the qcd matter is local and knows nothing of this global geometry .
therefore , to a first approximation the initial stress tensor will be locally azimuthally symmetric .
subsequently , if there are no reinteractions the produced matter will free stream to the detector ; the initial geometry will have no influence on the evolution , and the angular distribution of the final observed hadrons will also be azimuthally symmetric . on the other hand , if there are strong interactions which maintain local thermal equilibrium , the pressure gradients in the @xmath4 direction will be larger than in the @xmath5 direction , an anisotropy in the collective flow will develop , and ultimately an anisotropy in the momentum spectrum of the final hadrons will be observed .
the final momentum anisotropy is characterized experimentally by @xmath6 , the second harmonic of the azimuthal distribution of the produced particles with respect to the reaction plane .
experimentalists have measured @xmath6 as a function of transverse momentum @xmath3 , particle type , and impact parameter @xcite .
these results are surprisingly well described by ideal hydrodynamics @xcite , which amounts to the approximation that the interactions are fast enough to maintain the matter in equilibrium from an early time until hadronic freeze - out .
there are some limits to this success .
first , the measured @xmath6 falls below the ideal hydrodynamic prediction for momenta larger than @xmath7 .
second , the hydro fit fails to reproduce certain relative trends observed in the baryon and meson elliptic flows .
these trends are compactly summarized by
constituent quark scaling " @xcite which generally has been attributed to a kind of coalescence of constituent quarks @xcite . here
we will argue that the first corrections to equilibrium can clarify both of these shortcomings without the need for a coalescence model . to quantify the corrections to ideal hydrodynamics
it is important to study nonideal ( viscous ) hydrodynamics . in the last two years
there has been a major push in this direction @xcite .
these studies have used various formalisms and have studied variations of @xmath0 with respect to the input shear viscosity , the model for the initial geometry , and various other nuisance parameters .
however , we want to point out here that these studies have all made a common assumption about the way that the asymmetry in the stress tensor is manifested in the particle distribution after freezeout . in particular , the particle distribution after freezeout is locally of the form @xmath8 where @xmath9 is the equilibrium distribution and @xmath10 is the first correction .
all groups have assumed that @xmath11 and that the coefficient of proportionality is independent of particle type . in this paper
we will argue that this assumption matters , and that it is far from secure .
after an overview of the issue in the next section , in we will discuss the physics which establishes the momentum dependence of @xmath10 and its behavior in several theories .
we will see that while the most studied theories give @xmath12 , the most qcd - like theories do not .
then we explore the behavior of multi - component plasmas in .
we see there that the viscous corrections @xmath13 for different species are generically different .
this fact can account for the `` constituent quark scaling '' observed in the baryon and meson elliptic flows without any reference to the hadronization process .
we then make our concluding remarks .
some technical material is postponed to appendices . throughout
, we will denote 4-vectors with capital letters @xmath14 and use @xmath15 for their 3-vector components , @xmath16 for their energy components , and @xmath17 for @xmath18 .
our metric convention is [ ,+,+,+ ] , so that @xmath19 .
we use tilde to indicate momenta scaled by temperature , @xmath20 .
we will mostly write @xmath21 for the equilibrium distribution function , @xmath22 but will occasionally use @xmath23 when common convention dictates its use .
the appropriate statistics will be clear from context .
the energy momentum tensor is given by the sum of its ideal and dissipative parts in terms of four components of @xmath24 .
the other six independent components of @xmath24 can always be accommodated by a @xmath25 satisfying @xmath26 . ]
t^=(+)u^u^+ g^+^ , and obeys the equation of motion , _ t^=0 .
[ eq : eom ] in the first - order ( or navier - stokes ) approximation the dissipative part of the stress energy tensor in the local rest frame is ^ij=-(^i
u^j+^j u^i - ^ij_k u^k ) -^ij
-2^i u^j , [ eq : evol ] where @xmath27 is the shear viscosity , and we use @xmath28 to indicate that the bracketed tensor should be symmetrized and made traceless .
it is well known that the first order theory is plagued with difficulties such as causality violations and instabilities @xcite . in order to circumvent these issues a second order theory
is required .
the most commonly used second order relativistic viscous hydrodynamics is due to israel and stewart @xcite . for technical reasons we use a theory developed by ttinger and grmela @xcite .
the two theories are qualitatively the same ( _ i.e. _ for sufficiently small relaxation times they both approach the first order theory ) . to streamline the presentation
we postpone the details of our hydrodynamic model to and refer to previous work @xcite .
the solutions to the hydrodynamic equations yield the underlying temperature and flow profiles in the presence of viscosity .
particle spectra are then computed using the cooper - frye @xcite formula e=_f ( ) p^d _ , where @xmath29 and @xmath30 is the freeze - out hypersurface taken as a surface of constant energy density in this work . for a system out of equilibrium @xmath31
is not the equilibrium distribution function but also contains viscous corrections , f()=f_0()+f ( ) , where @xmath9 is the ideal bose / fermi distribution function .
the form of @xmath10 is constrained by the requirement that @xmath32 be continuous across the freeze - out hypersurface : t^ = p^p^f ( ) ^ij = p^i p^j f ( ) .
[ deltaf_and_pi ] dropping @xmath10 from the final particle spectra is inconsistent as it leads to a discontinuity in @xmath24 .
the form for @xmath10 which satisfies continuity in the local rest frame is proportional to @xmath33 and is traditionally parametrized by @xmath34 is treated as a dynamical variable in a second order fluid formalism .
then to first order one can make the replacement , @xmath35 .
there has been no attempt to systematically include @xmath10 through second order in hydrodynamic simulations . ]
[ chip ] f()&=&-n_p(1n_p ) ( ) , + & = & -(p ) n_p ( 1 n_p ) ^i ^j _
i u_j , where we have distinguished @xmath34 and @xmath36 by the argument of the function .
one moment of @xmath34 is fixed by the shear viscosity ( see below ) but otherwise @xmath34 is an arbitrary function of @xmath37 .
to date all works on viscous hydrodynamics have taken the quadratic and have usually worked in a boltzmann approximation ( p)p^2 . from a viscous hydrodynamic model employing the quadratic .
the run parameters are @xmath38 , @xmath39 mev and @xmath40 .
further details are in .
[ fig : quad ] ] let us look at typical results for @xmath0 as shown in .
the curve labeled ` ideal ' shows the result using ideal hydrodynamics ( _ i.e . _
the curve labeled ` @xmath9 ' shows the resulting elliptic flow from the viscous evolution ( the solution of eqs .
( [ eq : eom]),([eq : evol ] ) and ( [ eq : evol2 ] ) ) but without including the viscous correction to the distribution function .
in other words , this shows how the viscous correction to the temperature and flow profiles manifests itself in the particle spectra . only modest corrections to the spectra
are found . as already emphasized , this result is unphysical since dropping @xmath10 violates continuity of the stress tensor .
last , the curve labeled ` @xmath42 ' also takes @xmath10 into account , using the quadratic .
the viscous correction to the distribution function dominates the reduction in @xmath6 at large @xmath3 .
that means that the @xmath10 term is responsible for a significant part of the effects of viscosity in the particle spectra .
this being the case , it is imperative to perform a systematic study on the form of the viscous correction as well as its effect on elliptic flow . most of this paper will discuss the form of the viscous correction appearing in weakly - coupled qcd .
although this is not a theory of hadronizing qcd , it is one theory where quantitative first - principle calculations can be performed .
one of our major findings is that not all models of energy loss give the same predictions for the off - equilibrium distribution function .
in this section we consider a number of theories , to show that while the dependence @xmath43 is expected in some cases , other functional dependence is expected in others , including weakly coupled qcd and a hadron ( resonance ) gas .
the theories where we can make a definite statement about the functional form of @xmath10 are all described by kinetic theory . since freeze - out
is defined as the point where scatterings go from being common to being rare on the time scale of the evolution of the system , we generally expect that , _ just _ before freezeout , kinetic theory should be a reasonable description . within kinetic theory , the distribution function @xmath44 is determined by a boltzmann equation , [ boltzmann ] ( _ t + v_*p*_*x * ) f ( , ) = -c[f , ] , where @xmath45 $ ] is the collision operator . in equilibrium the distribution function obeys n ( , ) = , [ n , ] = 0 . to determine the first viscous correction @xmath10
, we work in a vicinity of the local rest frame @xmath46 , and substitute @xmath47 into keeping terms first order in the spatial derivatives [ linboltz ] n_p(1 n_p ) _
i u_j = -c[f , ] . here
$ ] denotes the linearized collision operator , @xmath49 the collision operator expanded to first order in @xmath10 . in writing
we have used ideal hydrodynamics and thermodynamic relations to rewrite time derivatives as spatial derivatives , and we have neglected gradients proportional @xmath50 which are responsible for the bulk viscosity @xcite . is an integral equation for @xmath10 which can be solved by various methods .
since the first viscous correction is a scalar and must be proportional to the the strains , the most general form for the viscous correction in the local rest frame can be parametrized by the function @xmath34 as in .
close to equilibrium the first viscous correction @xmath10 determines the strains ^ij = - 2 ^i u^j = f , which ultimately yields a relation between between the shear viscosity and the viscous correction @xmath34 [ constraint ] = n_p(1 n_p ) ( p ) .
this is the only general constraint on the functional form of the viscous distribution function . to proceed further
we must specify completely the form of the linearized collision operator , which we will do in the context of various model theories .
the simplest model ( really a cartoon ) for the collision operator is the relaxation time approximation , [ f , ] = , where @xmath51 is the momentum dependent relaxation time to be specified .
substituting this form for the collision operator into , and working in a boltzmann approximation @xmath52 yields the following form for @xmath10 : [ deltaf ] f = -n_p p^ip^j_i u_j .
note however that the relaxation time is in general energy dependent . in different theories
, @xmath53 might show different functional dependence on @xmath54 . without details about the dynamics of the theory in question , we can only parametrize the viscous correction . here
we will discuss a massless classical gas where @xmath55 and parameterize the relaxation time ( or the distribution function ) with a simple power law f(p)&=&-n_p()^i ^j , + ( ) & = & c()^2- .
[ deltaf2 ] the constant , @xmath56 , is determined through : c ( ) = .
there are two limiting cases for the functional form of the the relaxation time approximation , @xmath57 and @xmath58 .
the momentum dependence of the relaxation time in these extreme cases is _
r(p ) \ { rl p & = 0 + & = 1 . most theories will lie between these two extreme limits .
, since the cross section grows rapidly with energy , @xmath59 .
] loosely speaking , if the energy loss of high momentum particles grows linearly with momentum , @xmath60 one expects a relaxation time independent of momentum , @xmath61 . on the other hand
if the energy loss approaches a constant @xmath62 , the relaxation time will grow with the particle momentum @xmath63 .
shows the elliptic flow computed using these two functional forms for the first viscous correction .
it is important to emphasize that shear viscosity is the same in both cases . examining these figures ,
we see that the integrated elliptic flow is largely insensitive to the functional form of the first viscous correction .
this is because the integrated @xmath6 is primarily determined by the hydrodynamic variables @xmath64,@xmath25 which are independent of the functional dependence of the relaxation time @xcite .
the differential elliptic flow @xmath0 is sensitive to the rate of equilibration especially above @xmath65 . using the linear or quadratic for the distribution function .
right : integrated @xmath6 versus centrality showing independence from the precise form of the viscous correction .
run parameters can be found in .,title="fig:",scaledwidth=49.0% ] using the linear or quadratic for the distribution function .
right : integrated @xmath6 versus centrality showing independence from the precise form of the viscous correction .
run parameters can be found in .,title="fig:",scaledwidth=49.0% ] scalar field theory has been described at length by jeon @xcite , who rigorously derived the boltzmann equation and its collision kernel and then solved for @xmath67 numerically .
but if we make the approximation of boltzmann statistics , we can actually solve for @xmath10 in closed form .
first the non - linear boltzmann equation with bose - einstein statistics is c[f,]= _ , , _ , where the transition rate ( including a final state symmetry factor ) is [ eq : gamma ] _ = ( 2)^4 ^4(p + k - p - k ) , and we have used the traditional short hand , @xmath68 . after linearizing , @xmath69 the linearized collision integral
is [ 22cofdf ] _
i u_j = _ , , _
n_p n_k ( 1 + n_p)(1+n_k ) . at this point
we will make the boltzmann assumption by neglecting the stimulation factors , @xmath70 , and using , @xmath55
. then we will try a solution of the form [ eq : phiquad ] ( ) = c p^i
p^j _ i u_j /t^3 , @xmath49 assuming that @xmath71 or @xmath57 . substituting this form into the integral equation ( and ) , and performing the integrals yields ( see ) [ scalar_res ] _
i u_j & = & c _
i u_j p_i p_j .
thus , taking @xmath72 , the quadratic form in has provided an exact solution solution to the linearized integral equation .
the viscosity is @xmath73 in a boltzmann approximation .
physically , this happens because of the form of the scattering cross - section . since @xmath74 and @xmath75 , the cross - section scales as the inverse of the particle s energy .
the typical scattering is nearly randomizing , but high energy particles undergo fewer scatterings than low - energy ones .
therefore we find the same functional form as for momentum diffusion but for very different reasons .
this example from scalar field theory , together with the example of momentum diffusion ( see below ) , are the reason that most people assume the quadratic ansatz , @xmath43 should hold . in this section we will use the boltzmann equation for pure - glue qcd in three approximation schemes to calculate the first viscous correction . first we will consider a leading @xmath76 approximation where the dynamics can be summarized by a fokker - planck equation which describes the momentum diffusion of quasi - particles . in this limit we will find that the viscous correction is quadratic at large momentum , @xmath43 . next we will consider the qcd boltzmann equation but consider only @xmath77 collisions and neglect collinear radiation . in this limit , we will find that the viscous correction at large momentum behaves as @xmath78 .
finally , we will also include collinear radiation in the boltzmann equation as is necessary in a complete leading order treatment @xcite .
we will find that collinear radiation controls the relaxation of the high momentum modes and asymptotically we have @xmath79 , where the coefficient of proportionality is set by the rate of transverse momentum broadening , @xmath80 .
the impatient reader may skip to which summarizes the results of these three approximation schemes . in a leading log approximation , @xmath81 is considered a large number and the dynamics describes soft coulomb scattering .
each soft collision involves a small momentum transfer of order @xmath82 , but these collisions happen relatively frequently at a rate of @xmath83 ( neglecting logarithms ) .
thus a typical particle with momentum @xmath84 will diffuse in momentum space and equilibrate on a time scale of @xmath85 .
the resulting boltzmann equation linearized around equilibrium can be written as a fokker - planck equation @xcite [ leading_log ] _ t f + v__x f = t ( n_p(1 + n_p ) ) + , where @xmath86 is the drag coefficient of a high momentum gluon in this approximation scheme@xcite [ bthoma_soft ] = - , = t^2 ( ) . the precise form of the gain terms has been given in @xcite , but only involves the @xmath87 spherical harmonic components of @xmath88 , @xmath49 @xmath89 and @xmath90 . in the hydrodynamic limit considered here
@xmath88 is proportional to a traceless rank 2 tensor ( @xmath91 ) and these gain terms vanish . substituting the form of into leads to the following equation for @xmath34 : [ ll_fokker ] n_p ( 1 + n_p ) = tn_p ( 1 + n_p ) ( - + ( - ) + ) ( p ) .
we are not aware of a closed form solution to this equation , but we can find a solution for @xmath34 at large momentum . making the approximation @xmath92 , we find that ( p)= solves this equation .
this is the well known _
quadratic _ .
we next will consider the qcd boltzmann equation but we will neglect collinear radiation . we emphasize that this is not a consistent approximation scheme .
nevertheless , it illustrates clearly the relative roles of hard collisions and inelastic processes in determining the functional form of @xmath34 in the relevant sub - asymptotic regime .
the linearized boltzmann equation is the same as , but the squared matrix element is [ megg ] | m |^2= 8g^4 c_a^2 ( 3 - -- ) , which describes @xmath93 gluon scattering after summing over all spins and colors and dividing by the gluon degeneracy factor @xmath94 .
these matrix elements must be dynamically screened using hard thermal loops .
a procedure which is consistent at leading order ( where the debye mass is small ) but which makes a reasonable estimate when the debye mass is not small has also been described in @xcite , and we can follow exactly the numerical procedure of that reference to find @xmath95 .
we can also study the asymptotic behavior more directly . at asymptotically large momentum where
@xmath96 may be considered large , shows that ( p ) .
the constant in front of the log is related to @xmath97 , the rate of collisional energy loss of a gluon with momentum @xmath37 , [ chi_coll ] ( p ) = . in a leading @xmath98 approximation
the loss rate is @xcite [ bthoma ] _
p = t^2 ( ) , as is rederived in .
the above asymptotic form agrees well with the numerical solution of the boltzmann equation .
early calculations of the shear viscosity in pure - glue qcd found @xmath71 , that is , @xmath57 @xcite .
however this is because they were leading - log treatments , which reduced to momentum diffusion discussed above .
it was realized in @xcite that inelastic number changing processes are only suppressed by a log , but are enhanced at large energy @xmath99 by a factor of @xmath100 and dominate equilibration for @xmath101 .
this should not be a surprise .
after all , if we think about `` equilibration '' ( energy loss ) in qed , we find that although the leading order mechanism for the energy loss of a high energy electron is ionization ( elastic scattering ) , bremsstrahlung actually dominates the loss rate .
this is the case because in bremsstrahlung the energy lost per scattering can scale with the incident energy , rather than being incident energy independent as is the case with ionization . as a result ,
the penetration depth of an electromagnetic shower scales only logarithmically with the incident energy , _
i.e. _ the relaxation time is constant up to logs , @xmath102 .
if the same behavior occurred in qcd we would expect the linear to hold , @xmath58 .
the current understanding of energy loss in perturbative qcd is that the high energy behavior lies between these extremes .
high - energy particles in a qcd plasma lose energy predominantly by inelastic gluon radiation and the time scale for energy loss is short compared to the time scale for momentum diffusion ( `` jet broadening '' ) . in particular it was shown by baier _
et al _ that for @xmath103 the rate of ( inelastic ) energy loss scales with the incident energy as @xmath104 , with the half - integer power arising from the lpm suppression @xcite .
this implies a `` relaxation time '' which scales as @xmath105 , and therefore @xmath106 @xcite .
let us see how this emerges in the behavior of pure - glue qcd .
the point is that the boltzmann equation for a gluon plasma possesses both an elastic scattering term and an inelastic effective @xmath107 scattering term , _
t f+v_*p*_*x * f=-^22[f ] -^12[f ] .
[ eq : bltz1 ] this equation was first solved at leading order in @xmath108 by arnold , moore and yaffe to determine the shear viscosity @xcite .
their approach involved writing a multi - parameter for @xmath95 in terms of a basis of test functions .
while the determination of @xmath27 improves quadratically with the test function basis , the determination of @xmath95 improves only linearly . therefore to get good accuracy out to @xmath109 requires the use of a large basis of functions .
we find a basis of eight functions is sufficient and the @xmath110 behavior is already clear with such a basis .
defined in eq.(2.32 ) of the reference by @xmath111 . ]
we can also directly establish the asymptotic form of the solution . at asymptotically high momentum near collinear
dominates the equilibration of gluons .
we therefore look at the boltzmann equation including only @xmath112 splittings , _
t f+v_*p*_*x * f=-^12[f ] .
the relevant collision integral for near collinear joining and splitting of gluons at leading order in @xmath108 was worked out in @xcite : ^12=_0^dpdk(-p - k)(*p*;p , k ) , and is given in terms of the splitting function for @xmath113 .
in general this splitting function involves the solution of an integral equation which includes the lpm effect .
however , in the _ deep _ lpm regime @xcite where @xmath114 can be treated as small , the following leading log result for the splitting function can be obtained , ^g_gg(p;xp,(1-x)p)= .
the above splitting function contains the transport parameter @xmath80 , which characterizes the typical transverse momentum squared transferred to the particle per unit length . with the above splitting function
we show in that the solution of the off - equilibrium distribution function at asymptotically large momentum is _ g(p)p^3/2 .
let us now summarize some of the main features of the off - equilibrium dynamics of pure glue qcd at weak coupling . in the previous three sections we looked at the behavior of the off - equilibrium correction for pure glue qcd in various approximation schemes , deriving asymptotic behavior in each case .
these asymptotics are listed in . in this section
we wish to focus on the phenomenologically more interesting region where the equilibrating parton has intermediate energies ( @xmath115 ) . in this case
one must resort to numerical solutions of the boltzmann equation which we present in . to summarize
, we will discuss the curves from top to bottom starting with the quadratic " curve . in the leading @xmath76 approximation the linearized boltzmann equation simplifies to a differential equation , . the numerical solution to this
has been worked out in @xcite and is well described for all momenta by the asymptotic quadratic form , @xmath116 .
the numerical result will be presented in a forthcoming work @xcite , and for now we show the quadratic result as the solid blue line ( color online ) . next we considered qcd with the @xmath117 gluon scattering matrix element at leading order . the agreement between the asymptotics derived in the previous section and the numerical solution
can be found in . at intermediate momentum
we show the numerical solution of the boltzmann equation without inelastic processes as the data points under the curve labeled `` coll . '' .
the solid curve is the result of a power law fit at intermediate momentum , @xmath118 . in the leading order ( lo ) treatment
when is included , we find further equilibration of the gluons and our numerical results are reasonably described by the fit , @xmath119 .
finally , the linear ansatz is also shown in for comparison .
processes ( _ i.e. _ collisional energy loss only ) .
the lines are @xmath120 for @xmath121 going from top to bottom.,scaledwidth=55.0% ] one can now ask how the observed @xmath119 of pure glue at leading order will affect the viscous corrections to elliptic flow .
first of all , as we have already shown , the integrated @xmath6 will change marginally .
the differential @xmath6 , on the other hand , will be largely affected at higher @xmath3 .
this result is shown in along with the quadratic and linear for comparison . for a perturbative gluon gas at leading order . the linear and quadratic
are shown for comparison .
run parameters can be found in .,scaledwidth=55.0% ] the above considerations have shown that the relaxation of the high energy tail of the distribution is largely controlled by energy loss .
the low / intermediate momentum region is constrained by the shear viscosity via .
the _ strength _ of the off equilibrium correction is controlled by two non - perturbative parameters : @xmath27 at low momentum and @xmath80 at high momentum .
this is clearly seen by looking at the forms of @xmath122 we have found for pure glue qcd at leading order , ( p)= \ { rl ^1.38 & 5p 10 + p^1.5 & ^-1()1 . .
in we show plots of @xmath122 for various choices of the non - perturbative parameters @xmath123 and @xmath124 . the main point to take away
is the need for a consistency between @xmath27 and @xmath80 such that the low and high momentum regions of @xmath122 can merge smoothly into one another .
the three values of @xmath125 we have chosen reproduce the experimentally observed @xmath126 @xcite when convoluted with the higher twist @xcite , amy @xcite and asw @xcite energy loss models respectively .
it appears to be difficult to reconcile the discontinuity of @xmath122 between the lowest shear viscosity @xmath38 and smallest value of @xmath80 used in modeling heavy ion collisions . for a perturbative gluon gas at leading order for three values of the non - perturbative parameter @xmath123 .
the curves at _ higher _ momentum show the asymptotic forms of @xmath122 for three values of the non - perturbative parameter @xmath80 .
there must be a consistency between @xmath123 and @xmath80 in order that the curves merge at intermediate momentum.,scaledwidth=55.0% ] one might also ask what scattering behavior is expected at lower temperatures , in a hadron gas . how do the highest energy hadrons equilibrate , as a function of hadron energy ? a complete study requires understanding the energy - dependent hadron - hadron cross section , which has nontrivial energy dependence and must be determined from experiment
. however we should be able to say something about the high momentum behavior .
in hadron - hadron scattering , the inelastic branching fraction rises with increasing @xmath127 , dominating the cross - section for kinetic energies well above @xmath128 .
since generically no daughter in an inelastic collision carries more than half the energy of the initial high @xmath37 particle , we can take scatterings to be momentum randomizing ( the relaxation time approximation is sensible ) , especially for the highest energy hadrons . the relaxation time is then controlled by the scattering rate , @xmath129 , with @xmath130 the hadron number density and @xmath131 an averaged total hadronic cross - section . so
what is the behavior of the total hadronic cross - section ? at low momenta it is complicated by resonances but at large momenta there is universally a rising total cross - section .
therefore the relaxation time @xmath132 should naively involve a small or zero power of @xmath99 , that is , @xmath133 is expected , at least for the very high energy tail .
suggests @xmath134 . ]
certainly we do not expect @xmath57 .
however any more detailed discussion must be either model or data driven and lies outside the scope of this paper .
the plasmas just considered are treated as single - component , in the sense that all degrees of freedom are related to each other by symmetries ( spins by parity , colors by gauge invariance ) .
the quark - gluon plasma is a multi - component plasma . treating @xmath135 as small and @xmath136 as large
, the three light quark types behave the same , but the gluons behave differently from the quarks .
similarly , the hadronic plasma present at lower temperatures contains both baryons and mesons , each of several types .
the different components generically have different departures from equilibrium , that is , @xmath137 , which would manifest as different viscous corrections to their @xmath3 spectra . in particular , we will argue that faster equilibration for baryons than for mesons can give a simple explanation for the `` constituent quark scaling ''
@xcite observed in @xmath0 for mesons and baryons , without invoking any model of coalescence .
we now consider a two component gas of quarks and gluons and label the distribution functions with subscripts @xmath138 and @xmath139 respectively : f_g(p)&=&-n_p(1+n_p)_g()^i ^j , + f_q(p)&=&-n_p(1-n_p)_q()^i ^j . for use in hydrodynamic simulations we will again fit the off - equilibrium component of the quarks and gluons distribution function to the following power law , _
g()&=&c_g(_g)^2-_g , + _ q()&=&c_q(_q)^2-_q .
the results of the numerical solution of the boltzmann equation for the two component case are shown as points in .
the solid curves are the results of the fit done at intermediate momentum ( @xmath140 ) with the result @xmath141 . .
[ fig : chi2comp ] ] in order to solve for the two constants ( @xmath142 and @xmath143 ) we need two constrains .
the first constraint relates the coefficients @xmath144 to the shear viscosity , = _ a = q , g_a c_a p^3-_a n_p(1n_p ) . the sum is over quarks and gluons with degeneracies @xmath145 and @xmath146 .
the second constraint comes from fixing the ratio of @xmath147 to the numerical solution of the boltzmann equation .
this ratio is shown in and at large enough momentum ( @xmath148 ) we find 1.70 .
the explicit computation of the two coefficients ( @xmath144 ) in terms of the above ratio and @xmath123 is worked out in . in
we show the elliptic flow of quarks and gluons .
note the larger suppression for quarks as the gluons are forced into equilibrium much quicker .
this quicker relaxation can not simply be explained by naively assuming casimir scaling , @xmath149 .
instead this ratio involves a playoff between the faster equilibration rate of gluons and the tendency of identity changing processes @xmath150 , @xmath151 , @xmath152 to equilibrate disequilibrium between the quarks and gluons .
this ratio is evaluated analytically at asymptotically large momentum in .
the distinct quark and gluon elliptic flow is completely due to the different viscous corrections , which in turn is related to the different relaxation rates of quarks and gluons .
let us note that if we scale both the @xmath6 and @xmath3 of gluons by three and quarks by two , the result is a `` universal curve '' as shown in the right plot of .
the observed scaling is completely accidental , but it led us to consider the possibility of finding similar scaling behavior in a meson / baryon system due to differences in the relaxation rates .
this is discussed in detail in the next section . and @xmath3 scaled by n=3,2 for gluons and quarks respectively .
run parameters can be found in .,title="fig:",scaledwidth=49.0% ] and @xmath3 scaled by n=3,2 for gluons and quarks respectively .
run parameters can be found in .,title="fig:",scaledwidth=49.0% ] mesons and @xmath153 baryons from viscous hydrodynamics with radiative or quadratic .
the run parameters are @xmath154 , @xmath155 mev and lattice eos .
further details are in .
the data is from the star collaboration @xcite .
[ fig : v2pipradquad],title="fig : " ] mesons and @xmath153 baryons from viscous hydrodynamics with radiative or quadratic .
the run parameters are @xmath154 , @xmath155 mev and lattice eos .
further details are in .
the data is from the star collaboration @xcite .
[ fig : v2pipradquad],title="fig : " ] mesons and @xmath153 baryons from viscous hydrodynamics with radiative or quadratic .
the run parameters are @xmath154 , @xmath155 mev and lattice eos .
further details are in .
the data is from the star collaboration @xcite .
[ fig : v2pipradquad],title="fig : " ] mesons and @xmath153 baryons from viscous hydrodynamics with radiative or quadratic .
the run parameters are @xmath154 , @xmath155 mev and lattice eos .
further details are in .
the data is from the star collaboration @xcite .
[ fig : v2pipradquad],title="fig : " ] the qcd matter immediately before freezeout is certainly not a weakly coupled quark - gluon plasma , but it might be described as a hadron ( resonance ) gas . just as for the quarks and gluons , there is no reason to think that the mesons and baryons should show the same efficiency in equilibrating .
but rather than claim a specific model for the partially equilibrated state of such a system , we will just do some phenomenology to see how different thermalization rates could affect the observed species - dependent elliptic flow behavior . to study the hydrodynamics of this system we switch from the ideal gas equation of state to a lattice motivated @xcite .
further details of the simulation are presented in .
we consider a meson / baryon gas whereby mesons and baryons have the off - equilibrium corrections @xmath156 and @xmath157 respectively , f_m(p)&=&-n_p(1+n_p)_m()^i ^j , + f_b(p)&=&-n_p(1-n_p)_b()^i ^j . we assume both species have the same power - law correction to spectra , _
m()&=&c_m()^2- , + _ b()&=&c_b()^2- , but we allow for different coefficients ( @xmath158 ) which we will choose in order to give reasonable agreement with data . for simplicity , we will consider two different : quadratic ( @xmath57 ) and radiative ( @xmath159 ) , and take the following ratios which , as we will show , fit the data rather well : = \ { cc 1.6 & + 1.4 & .
finally , the numerical values of the coefficients can be identified with the shear viscosity through = _
a=,k, ... _a
c_m / b p^4- n(e_a ) , where the sum extends over all mesons / baryons having @xmath160 gev respectively .
this choice reproduces the lattice parametrization of the equation of state below @xmath161 mev .
we find the following values for the coefficients at our freeze - out temperature of @xmath162 mev , & & .
c c_m=1.053 + c_b=0.658 + } ( ) + & & .
c c_m=2.661 + c_b=1.901 + } ( ) before computing particle spectra we would like to make an aside about the way elliptic flow is computed . by definition @xmath0
is given by v_2(p_t ) , [ eq : v2def ] where @xmath163 is short for @xmath164 $ ] and @xmath165 is the first viscous correction to this . in the above expression
the viscous correction to the phase space distribution , @xmath166 , occurs both in the numerator as well as in the normalization from the denominator .
since we have restricted the viscous correction to be linear in gradients of field quantities we should therefore require that @xmath6 be computed to the same order .
we therefore expand the denominator v_2 - , [ eq : v2expand ] so the expression retains terms to first order in @xmath10 only . in the following
we will show both the expanded and unexpanded expressions for @xmath6 , shading the region between the two results in order to give an estimate for the uncertainty in the gradient expansion .
the upper limit of the band corresponds to while the lower limit is . in figures where
the uncertainty band is omitted the plotted curve corresponds to .
let us now discuss how the different fare with the experimental data .
we have chosen @xmath154 in order to give reasonable agreement with the data in the transverse momentum range @xmath167 } \leq 2 $ ] .
the @xmath0 spectra for @xmath168 and @xmath153 are presented in using either the radiative ( @xmath169 ) or quadratic ( @xmath1 ) . for @xmath170 gev
we find good agreement between the viscous hydrodynamic results and data .
if one included hadronic rescattering the low momentum component of the @xmath153 @xmath6 would be pushed out towards higher @xmath3 giving better agreement with the data . above 2 - 3
gev large differences between the radiative and quadratic are realized .
we must warn that at higher @xmath3 one can not make a direct comparison with data since a larger fraction of the yield will come from fragmenting partons , which have not been included .
in addition , the hydrodynamic description starts to break down at larger @xmath3 .
regardless , one must keep in mind that for large enough momentum ( _ i.e. _ @xmath171 gev ) the two used here are clearly discernible and the choice of could in principle lead to differences in the extracted viscosity . of @xmath168 and @xmath153 .
right : constituent quark scaling of @xmath0 .
run parameters can be found in .
the data is from the star collaboration @xcite and is plotted in @xmath172 as suggested by the phenix collaboration @xcite .
( recent phenix data presented at quark matter @xcite clearly deviate from constituent quark scaling above @xmath173 gev . )
[ fig : v2pipscaled ] , title="fig : " ] of @xmath168 and @xmath153 .
right : constituent quark scaling of @xmath0 .
run parameters can be found in .
the data is from the star collaboration @xcite and is plotted in @xmath172 as suggested by the phenix collaboration @xcite .
( recent phenix data presented at quark matter @xcite clearly deviate from constituent quark scaling above @xmath173 gev . )
[ fig : v2pipscaled ] , title="fig : " ] of @xmath168 and @xmath153 .
right : constituent quark scaling of @xmath0 .
run parameters can be found in .
the data is from the star collaboration @xcite and is plotted in @xmath172 as suggested by the phenix collaboration @xcite .
( recent phenix data presented at quark matter @xcite clearly deviate from constituent quark scaling above @xmath173 gev . )
[ fig : v2pipscaled ] , title="fig : " ] of @xmath168 and @xmath153 .
right : constituent quark scaling of @xmath0 .
run parameters can be found in .
the data is from the star collaboration @xcite and is plotted in @xmath172 as suggested by the phenix collaboration @xcite .
( recent phenix data presented at quark matter @xcite clearly deviate from constituent quark scaling above @xmath173 gev . )
[ fig : v2pipscaled ] , title="fig : " ] we would now like to investigate whether we observe a meson / baryon scaling , similar to the accidental quark / gluon scaling we found from first principles earlier . for clarity , we again present the above results with mesons and baryons on the same figure .
this is shown for both radiative and quadratic in .
the figures show the corresponding results with both @xmath6 and @xmath3 re - scaled by the number of constituent quarks .
the scaling of the data is the well - known phenomenon of constituent quark scaling .
we find that viscous hydrodynamics reproduces this `` universal curve '' as well .
this is due to the difference in relaxation rates between mesons and baryons , which was treated as a free parameter .
the possible microscopic origin of this ratio is discussed further in .
we should also point out that this relaxation time scaling is fairly robust to changes in the equation of state . while changing the equation of state will clearly affect the @xmath174 behavior
, these changes will only have modest modifications to the viscous correction to the distribution functions . the qualitative feature that species with smaller relaxation times have a stronger elliptic flow is borne out by ( a quark gluon plasma equation of state ) and ( a lattice equation of state ) .
further study of the equation of state is left to future work .
in this work we have presented a systematic study of the first viscous correction to the thermal distribution function .
all simulations of viscous hydrodynamics so far have used the quadratic ( p ) p^2 , but this is only an educated guess .
first we studied the form of @xmath10 ( or @xmath34 ) in a momentum dependent relaxation time approximation and derived a simple formula ( p ) = _ r(p ) . examining this formula we considered two special cases @xmath63 ( where the equilibration time is proportional to energy ) and @xmath175 ( where the equilibration time is independent of energy ) .
these give rise to quadratic ( @xmath43 ) and linear ( @xmath176 ) dependence on momentum , as is summarized in .
we expect that , provided qcd is describable in terms of quasi - particles , the first viscous correction should lie between these cases .
compares these two extreme limits for the functional form of the viscous correction .
it is important to emphasize that the two simulations have precisely the same shear viscosity . comparing our results for the elliptic flow in these two theories
, we see that the integrated elliptic flow @xmath6 is largely determined by the shear viscosity , while differential quantities such as @xmath0 at high @xmath3 depend on the equilibration rates at high momentum .
the integrated elliptic flow is determined to a large extent by the hydrodynamic variables @xmath177 .
( an explicit formula relating @xmath6 to @xmath64 and @xmath25 is given in which in turn was motivated by earlier observations @xcite . )
c @ | @ x @ | @ l model & physics & formula + relaxation time , @xmath63 & relaxation time grows with particle momentum . &
@xmath178 + relaxation time , @xmath179 & relaxation time independent of momentum .
& @xmath180 + scalar theory & randomizing collisions which happen rarely & @xmath43 + qcd soft scatt . & soft @xmath181 collisions lead to a random walk of hard particles . & @xmath182 + qcd hard scatt . & hard @xmath183 collisions lead to a random walk of hard particles . & @xmath184 + qcd rad .
e - loss & radiative energy controls the approach to equilibrium . in the lpm regime
@xmath80 controls the radiation rate . &
@xmath185 + the quadratic ansatz is valid only for fairly specialized theories . for instance , examining we see that scalar theories follow this .
the reason is that the cross - section falls as @xmath186 , so higher - energy particles see a more transparent medium and equilibrate more slowly . for different reasons the quadratic is also valid in a soft scattering approximation to high temperature qcd ( see row 4 of ) . in this limit , which treats @xmath76 as an expansion parameter , soft @xmath187 collisions
lead to the momentum diffusion and drag of hard gluons .
if the momentum diffusion is independent of particle energy and the drag is constant , we get the quadratic . if the momentum diffusion increases logarithmically with particle energy , we find a logarithmic correction to this ( see row 5 of ) . a formula which summarizes the asymptotic form of both of these cases is ( p ) = , where @xmath97 is the rate is energy loss of a particle with momentum @xmath37 ( see and for explicit formulas in certain limits ) .
however , the effect of bremsstrahlung completely changes this picture .
a naive ( bethe - heitler ) treatment of radiative energy loss would lead to a relaxation rate independent of momentum , but including the lpm effect , the viscous correction behaves asymptotically as [ qhat_chi_conc ] ( p ) = 0.7 .
this formula is summarized in row 6 of and provides a concrete connection between viscous corrections and radiative energy loss which is further explored in and surrounding text . from a phenomenological perspective , the lpm effect is not entirely dominant and collisions are important in the relevant momentum range .
a phenomenological fit to numerical results for the first viscous correction , including both collisions and collinear radiation without making the strict lpm approximation , shows that the first viscous correction is reasonably well described by the following phenomenological form : ( p ) c p^1.38 .
compares this functional form to the linear and quadratic motivated by the relaxation time approximation .
we see that the general expectation from high temperature qcd is that in the relevant momentum range the first viscous correction is slightly closer to the linear rather than the quadratic ansatz .
we next studied two component plasma starting with a two component plasma of quarks and gluons .
since the relaxation rates of the quarks and gluons are not the same the two components do not have the same distribution function . at high momentum
an analysis of collinear splittings @xmath188 , @xmath189 , @xmath190 shows that both the quark and gluon distribution behave as @xmath191
. however the ratio of the quark and gluon viscous corrections approaches a constant [ quark_glue_17 ] 1.70 .
the constant is determined by the ratio of casimirs @xmath192 and the dynamics of the qcd splitting functions .
it also depends weakly on the number of quark flavors and we have quoted the two flavor case . motivated by this example
, we have postulated that the baryon and meson components of the medium have different equilibration rates .
indeed , there is no reason to expect that these species would equilibrate at the same rate .
then we fitted ( by eye ) the ratio of relaxation rates to reproduce the baryon and meson elliptic flows .
if the ratio of relaxation times is 1.5 , meaning that baryons relax to equilibrium 1.5 times faster than mesons , then the resulting viscous hydrodynamic calculation effortlessly reproduces the universal `` constituent quark scaling '' curve .
physically what is happening is that in ideal hydrodynamics the baryons and mesons have approximately the same elliptic flow which is approximately described by a linear rise in @xmath193 .
the viscous correction then dictates that the baryons will follow this ideal trend 1.5 times farther than the mesons .
although it is not obvious from the data shown in , the data do not show scaling above @xmath194 , _
i.e. _ the last lambda point is a fluctuation upward ( this is seen quite clearly in recent phenix data@xcite . ) it is interesting that the data also deviate from hydrodynamic predictions above this point .
it is tempting to speculate as to the microscopic origin of the factor of @xmath195 .
the baryons and mesons in the @xmath196 region are produced in the complex transition region where the energy density decreases from @xmath197 to @xmath198 . in this range ,
the temperature decreases by only @xmath199 .
however , the hydrodynamic simulations evolve this complicated region for a significant period of time , @xmath200 , and the hadronic currents are built up over this time period .
the interactions are probably quite inelastic and are not easily classified as hadronic or partonic in nature .
the additive quark model was used to describe high energy total cross sections which are similarly inelastic @xcite .
it predicts the ratio of high energy nucleon - nucleon to pion - nucleon ( as well as pion - nucleon to pion - pion ) cross sections to be @xmath201 in reasonable agreement with the experimental ratio .
perhaps similar physics is responsible for the different baryon and meson elliptic flows .
in fact , the splitting of the baryonic and mesonic elliptic flows was predicted at least qualitatively by urqmd which implements the additive quark model @xcite . on the other hand ,
the factor of @xmath195 in the relative relaxation times could be simply a combination of dynamical and group theoretical factors of accidental significance . in summary , a species dependent relaxation time provides a coherent and physically transparent explanation for the complicated trends observed in the elliptic flow data measured at rhic .
kd is supported by the us - doe grant de - ac02 - 98ch10886 .
dt would like to thank paul sorenson , raimond snellings , and jiangyong jia for informative discussions .
dt is supported in part by an oji grant from the us department of energy de - fg-02 - 08er4154 and the sloan foundation .
gm would like to thank paul romatschke for useful conversations , the physics department at the universidad autonoma madrid for hospitality while this work was completed , and the alexander von humboldt foundation for its support through a f. b. bessel prize .
gm s work was supported in part by the natural sciences and engineering research council of canada .
the initial condition of the hydrodynamic evolution is set by a glauber model and the energy density is proportional to the number of binary collisions . more specifically we take ( _ 0=1,x , y)= e_bc where @xmath202 is the energy per binary collision and @xmath203 mb is the inelastic nucleon - nucleon cross section . in this work
we will use the following evolution equation for @xmath204 , @xmath205 which is identical to the stress tensor used in @xcite .
other possibilities are also possible @xcite which will not change the results of this work on a qualitative level . in the above expression @xmath206
is the vorticity and @xmath207 .
there is one technical detail that warrants discussion . at large transverse distances
the viscous pressure tends to become larger than the ideal pressure and the equations become unstable .
it is therefore necessary to cutoff our auxiliary tensor when it becomes large .
more precisely we take ^ij , where @xmath208 and @xmath209 . in the first part of this paper we consider an ideal gas equation of state , @xmath40 . for a two flavor qgp
the ideal stefan boltzmann gas gives @xmath210 , which roughly corresponds to the @xmath211 relation found on the lattice .
( for the highest temperatures in the simulation it is above this value and for the lowest temperatures in the simulation it is this value ) .
we have decided to use the same @xmath211 ratio for both the gluon gas and quark + glue simulations in order to get the fairest possible phenomenological estimate for the size of the viscous corrections in a realistic heavy ion event . for simulations using the ideal gas eos
the freeze - out contour is taken at constant @xmath212 gev/@xmath213 corresponding to a temperature of 140 mev .
the default impact parameter is 7.6 fm and the shear viscosity to entropy ratio is @xmath38 . in the second part of this paper where we compute spectra of a meson / baryon gas
we have used a lattice motivated equation of state @xcite . in this case
the freeze - out surface is set by @xmath214 gev/@xmath213 corresponding to a temperature of 150 mev .
we have used a default impact parameter of 6.8 fm corresponding to a centrality class of 10 - 40% and a shear viscosity to entropy ratio of @xmath154 .
in this section we will give the details leading to for a scalar theory and for pure glue .
our starting point is . substituting the form specified in into this equation yields in a boltzmann approximation & = & e^-k / t ( ^2 ) ( ^2 ) + & & ^4 ( ( + ) - - ) ( __+ _ _ - _ _ - _ _ ) , where @xmath215 is the lorentz invariant extension
, we have @xmath216 in the local rest frame implicit here , we have @xmath217 and @xmath218 for @xmath219 . ] of @xmath220 . the integrals over @xmath221 and @xmath222
are lorentz covariant and can be performed by standard tricks ; the @xmath223 term can be factored out , leading to @xmath224 , while the integral over @xmath225 must return a rank-2 tensor depending only on @xmath226 .
there are only two such tensors , and contraction with @xmath227 and @xmath228 establishes that @xmath229 the integral equation becomes & = & e^-k / t ( p_i p_j + k_i k_j - ( p+k)_i ( p+k)_j ) , where we used that @xmath230 is traceless , @xmath231 . performing the @xmath232 angular integration in the plasma frame , the @xmath233 terms integrate to zero ; so does the @xmath234 term , because @xmath235 is traceless . performing the trivial radial integration , we find .
our goal here is to derive and .
our starting point is the collision integral with matrix elements given by _
i u_j = _ n_p n_k ( 1 + n_k ) ( 1 + n_p ) _ pkpk .
using the definition , @xmath236 , one can pull out the common factor , @xmath237 .
the remaining integral on the right hand side ( called @xmath238 ) must have the form @xmath239 since this is the only symmetric traceless tensor which can be constructed out of @xmath240 and @xmath241 .
straightforward analysis then shows that & = & _
n_p n_k(1 + n_k ) ( 1 + n_p )
_ pkpk + & & , [ eq : coll3 ] where for instance p_2 ( _ ) = ( ^i ^j - ^ij ) ( ^i ^j - ^ij ) , is the second legendre polynomial
. we will evaluate this integral in a leading @xmath242 approximation .
asymptotically , the momenta @xmath37 and @xmath243 are large , while @xmath232 and @xmath244 are of order the temperature is small . since the particles are identical
, there is also an equal contribution where @xmath245 is small , _
i.e. _ when @xmath37 and @xmath244 are large and @xmath243 and @xmath232 are of order @xmath84 .
our original definition of the transition rate includes a @xmath246 symmetry factor for the identical particle final state . to ease the discussion in this section
, we will simply drop the symmetry factor and neglect @xmath247-channel contribution . ] . in this limit
we can make the boltzmann approximation , @xmath248 , and can treat @xmath243 as close to @xmath37 . specifically we take _
pp = 1 + 1 , and then write ( p ) - ( p ) - . we also note that @xmath232 and @xmath244 are close to @xmath84 and therefore @xmath249 and @xmath250 are small
. then we can write as _
p , where the average energy loss rate for a particle with momentum @xmath37 is _
p = - _ _ n_k ( 1 + n_k ) , the energy loss is the transition rate weighted with the energy transfer .
the energy loss to leading @xmath242 has been determined by bjorken @xcite and braaten and thoma @xcite and reads [ lleloss ] _ p
one can verify that when terms suppressed by @xmath242 are dropped we have [ chicoll3 ] ( p ) = _ ~t^~p p p^2 . shows a fit based on which does a reasonable job in reproducing our @xmath251 numerical results at high momentum .
form , @xmath252 , with fit parameter @xmath253 .
, scaledwidth=55.0% ] for completeness we will rederive . in order to evaluate the phase space integrals over @xmath254 we use the t - channel parametrization " of . following the logic that leads from ( a.14 ) to ( a.21 ) of this work
, we write the phase space as [ phasespacei ] _ _ = _ 0^ dk _ 0 ^ 2 d_ps dq _
ps d|m|^2 , where the momentum transfer is @xmath255 and the energy transfer is @xmath256 .
the vector @xmath257 is taken along the @xmath258 axis and the vector @xmath240 lies in the @xmath259 plane .
the angle @xmath260 is the azimuthal angle of @xmath261 with respect to the @xmath262 plane .
the energy transfer and momentum transfer are restricted to the available phase space 0 & < & < k , + 0 & < & < p , which is also exhibited in .
approximation.,scaledwidth=48.0% ] using the definitions of the kinematic variables , the mandelstam invariants are t & = & -(p-p)^2 = q^2 - ^2 , + s & = & -(p+k)^2 = , + u & = & -t - s . to evaluate the collision integral , we are to substitute these expressions for the mandelstam invariants into the matrix elements and perform the integrals over the phase space . close inspection of the result of this procedure shows how @xmath242 comes about .
first , the logarithm comes from integrating over the phase space region where @xmath263 and @xmath264 as shown by the band shown in .
since the @xmath265 integral is over the interval , @xmath266 , the phase - space integral is approximately _ ps dq _
~t^~p dq , and we may neglect the stimulation factor , @xmath267 .
second , only the highest powers of @xmath265 and @xmath138 contribute to the ultraviolet logarithm .
the @xmath260 integrated matrix elements with these restrictions is _ 0 ^ 2 d & & . then the total total transition rate is _
p = - _ 0^ dk 2 k n_k _ ~t^~p
dq ( -q ) . performing the integral over @xmath232 , we arrive at the result quoted in .
in this appendix we derive the form of the viscous distribution function for asymptotically large momenta . in doing this
we will relate the high @xmath3 tail of the distribution function with the energy loss parameter @xmath80 .
the starting point is the boltzmann equation containing near collinear splitting processes .
we neglect @xmath268 processes as these will be sub - leading at large momenta . therefore _
f_a(*x*,*p*)=-^12_a[f ] , where ^12_a= _ 0^dpdk(-p - k ) ( * p*;p , k ) . in the above expression @xmath269 is the distribution function of species @xmath270 .
the degeneracy factor , @xmath271 is 16 for gluons and @xmath272 for quarks .
now linearize the collision integral ^12=_0^dpdk(p - p - k)(p;p , k)n_p(1n_p)(1n_k ) . doing the integral over @xmath273 and expanding out the lhs in the typical way we get n_p(1n_p)=-_0^dx ( p;xp,(1-x)p)n_p(1n_xp)(1n_(1-x)p ) .
then note at very high momentum ( 1-x ) .
let us now consider a two component plasma of quarks and gluons . using the @xmath274 we are left with [ eq : c6 ] & = & p^2- _ 0 ^ 1 dx ^g_gg(p;xp,(1-x)p ) + & & + p^2- _ 0 ^ 1 dx ^g_q(p;xp,(1-x)p ) , + & = & p^2- _ 0 ^ 1 dx ^q_gq(p;xp,(1-x)p ) .
the splitting functions at leading log order are ^g_gg(p;xp,(1-x)p)&= & , + ^g_q(p;xp,(1-x)p)&= & , + ^q_gq(p;xp,(1-x)p)&= & .
the @xmath275 behavior here together with the @xmath276 behavior explicitly on the rhs of must cancel the @xmath277 behavior on the lhs of .
this fixes @xmath278 or @xmath279 , so @xmath280 .
this proves the claim in the main text that the asymptotic behavior should be @xmath281 . for a gluon gas
we find c_g^-1= _ 0 ^ 1 dx , and _
g = p^3/2 . for a two - flavor quark - gluon gas
g = p^3/2 , + _ q = p^3/2 .
the ratio is = 1.657 , not too different from the ratio @xmath282 we found by fitting .
this ratio depends on @xmath283 . for 1 flavor
it is 1.702 , for 3 flavors it is 1.618 , and in the limit of infinite flavors it approaches 1.128 .
this diminishing ratio occurs because , at larger @xmath283 , more and more splitting processes are @xmath152 and @xmath151 , which equilibrate the numbers of quarks and gluons towards each other .
in this appendix we derive relationships between the off - equilibrium distribution function , @xmath122 and the shear viscosity of a two component system .
we consider a gas of bosons and fermions as it will have applications to a gas of quarks and gluons or a gas of mesons and baryons .
n_p(1-n_p)_f(p)^i ^j , + f_b(p)&=&-_b n_p(1+n_p)_b(p)^i ^j .
the off - equilibrium correction @xmath122 takes the form _
b(p)&=&c_b(t)p^2-_b , + _ f(p)&=&c_f(t)p^2-_f .
the goal is to find values of the coefficients @xmath284 and @xmath285 as a function of @xmath84 and @xmath123 .
first we define the partial viscosity of each species _ f = p_f(p)n_p , + _ b = p_b(p)n_p , [ eq : d3 ] which will yield a total viscosity of = _ f+_b . for massive particles
the phase space integrals must be done numerically , but for massless particles , integrating yields c_f(t ) & = & , + c_b(t ) & = & , with @xmath286 .
let us define @xmath287 as the ratio of the partial viscosities , .
making use of the relations & = & _ f+_b=(1+)_b , + s&=&s_f+s_b = v_f+v_b , we find c_f(t ) & = & ( ) , + c_b(t ) & = & ( ) .
[ eq:2cfinal ] j. adams _ et al . _ [ star collaboration ] , nucl .
phys . a * 757 * , 102 ( 2005 ) [ arxiv : nucl - ex/0501009 ] .
k. adcox _ et al . _
[ phenix collaboration ] , nucl .
phys . a * 757 * , 184 ( 2005 ) [ arxiv : nucl - ex/0410003 ] .
b. b. back _ et al .
_ , nucl .
phys . a * 757 * , 28 ( 2005 ) [ arxiv : nucl - ex/0410022 ] . i. arsene _ et al . _ [ brahms collaboration ] , nucl . phys . a * 757 * , 1 ( 2005 ) [ arxiv : nucl - ex/0410020 ] .
t. hirano and y. nara , nucl .
a * 743 * , 305 ( 2004 ) .
d. teaney , j. lauret and e. v. shuryak , phys .
lett . * 86 * , 4783 ( 2001 ) ; _ ibid _ arxiv : nucl - th/0110037 .
p. f. kolb , p. huovinen , u. w. heinz and h. heiselberg , phys .
b * 500 * , 232 ( 2001 ) .
p. huovinen , p. f. kolb , u. w. heinz , p. v. ruuskanen and s. a. voloshin , phys .
b * 503 * , 58 ( 2001 ) . c. nonaka and s. a. bass , phys .
c * 75 * , 014902 ( 2007 ) .
h. song and u. w. heinz , phys .
b * 658 * , 279 ( 2008 ) [ arxiv:0709.0742 [ nucl - th ] ] .
k. dusling and d. teaney , phys .
c * 77 * , 034905 ( 2008 ) [ arxiv:0710.5932 [ nucl - th ] ] .
p. huovinen and d. molnar , phys .
c * 79 * , 014906 ( 2009 ) [ arxiv:0808.0953 [ nucl - th ] ] .
h. song and u. w. heinz , phys .
c * 77 * , 064901 ( 2008 ) [ arxiv:0712.3715 [ nucl - th ] ] .
p. bozek , phys .
c * 77 * , 034911 ( 2008 ) [ arxiv:0712.3498 [ nucl - th ] ] . w. a. hiscock and l. lindblom , annals phys . *
151 * , 466 ( 1983 ) .
w. a. hiscock and l. lindblom , phys .
d * 31 * , 725 ( 1985 ) .
w. israel , ann .
* 100 * ( 1976 ) 310 ; w. israel and j.m .
stewart , phys .
lett . * 58a * ( 1976 ) 213 .
m. grmela , h.c .
ttinger , phys .
e * 56 * , 6620 ( 1997 ) .
ttinger , m. grmela , phys .
e * 56 * , 6633 ( 1997 ) .
ttinger , phys .
e * 57 * , 1416 ( 1993 ) .
h. c. ttinger , physica a * 254 * ( 1998 ) 433 - 450 . f. cooper and g. frye , phys .
d. * 10 * , 186 ( 1974 ) .
arnold , g. d. moore and l. g. yaffe , jhep * 0111 * , 057 ( 2001 ) [ arxiv : hep - ph/0109064 ] .
p. arnold , g. d. moore and l. g. yaffe , jhep * 0301 * , 030 ( 2003 ) [ arxiv : hep - ph/0209353 ] .
p. arnold , g. d. moore and l. g. yaffe , jhep * 0011 * , 001 ( 2000 ) [ arxiv : hep - ph/0010177 ]
. j. hong and d. teaney , in preparation .
g. baym , h. monien , c. j. pethick and d. g. ravenhall , phys .
lett . * 64 * , 1867 ( 1990 ) .
h. heiselberg , phys .
d * 49 * , 4739 ( 1994 ) [ arxiv : hep - ph/9401309 ] .
r. baier , y. l. dokshitzer , a. h. mueller , s. peigne and d. schiff , nucl .
b * 483 * , 291 ( 1997 ) [ arxiv : hep - ph/9607355 ] .
r. baier , y. l. dokshitzer , a. h. mueller , s. peigne and d. schiff , nucl .
b * 484 * , 265 ( 1997 ) [ arxiv : hep - ph/9608322 ] .
p. arnold , g. d. moore and l. g. yaffe , jhep * 0206 * , 030 ( 2002 ) [ arxiv : hep - ph/0204343 ] .
p. arnold and c. dogan , phys .
d * 78 * , 065008 ( 2008 ) [ arxiv:0804.3359 [ hep - ph ] ] . s. a. bass , c. gale , a. majumder , c. nonaka , g. y. qin , t. renk and j. ruppert , phys .
c * 79 * , 024901 ( 2009 ) [ arxiv:0808.0908 [ nucl - th ] ] .
x. f. guo and x. n. wang , phys .
lett . * 85 * , 3591 ( 2000 ) .
x. n. wang and x. f. guo , nucl .
a * 696 * , 788 ( 2001 ) .
b. w. zhang and x. n. wang , nucl .
a * 720 * , 429 ( 2003 ) .
a. majumder , e. wang and x. n. wang , phys .
lett . *
99 * , 152301 ( 2007 ) .
a. majumder and b. muller , phys .
c * 77 * , 054903 ( 2008 ) .
a. majumder , r. j. fries and b. muller , phys .
c * 77 * , 065209 ( 2008 ) .
b. g. zakharov , jetp lett . *
63 * , 952 ( 1996 ) , hep - ph/9607440 .
b. g. zakharov , jetp lett .
* 65 * , 615 ( 1997 ) , hep - ph/9704255 .
b. g. zakharov , phys .
nucl . * 61 * , 838 ( 1998 ) , hep - ph/9807540 . c. a. salgado and u. a. wiedemann , phys
* d68 * , 014008 ( 2003 ) , hep - ph/0302184 .
u. a. wiedemann , nucl .
* b582 * , 409 ( 2000 ) , hep - ph/0003021 .
u. a. wiedemann , nucl .
* b588 * , 303 ( 2000 ) , hep - ph/0005129 . c. a. salgado and u. a. wiedemann , phys .
* 89 * , 092303 ( 2002 ) , hep - ph/0204221 .
n. armesto , c. a. salgado , and u. a.wiedemann , phys .
lett . * 94 * , 022002 ( 2005 ) , hep - ph/0407018 . m. laine and y. schroder , phys .
d * 73 * , 085009 ( 2006 ) [ arxiv : hep - ph/0603048 ] .
arkadij taranenko for the phenix collaboration , presented at _
quark matter 2009 _ , knoxville , tennessee , march 30april 4 ( 2009 ) . | this work investigates the first correction to the equilibrium phase space distribution and its effects on spectra and elliptic flow in heavy ion collisions .
we show that the departure from equilibrium on the freezeout surface is the largest part of the viscous corrections to @xmath0 .
however , the momentum dependence of the departure from equilibrium is not known _ a priori _ , and it is probably not proportional to @xmath1 as has been assumed in hydrodynamic simulations . at high momentum in weakly coupled plasmas
it is determined by the rate of radiative energy loss and is proportional to @xmath2 .
the weaker @xmath3 dependence leads to straighter @xmath0 curves at the same value of viscosity .
further , the departure from equilibrium is generally species dependent .
a species dependent equilibration rate , with baryons equilibrating faster than mesons , can explain `` constituent quark scaling '' without invoking coalescence models . |
supernova ( sn ) 2002ap in m74 was discovered by y. hirose and confirmed by r. kushida and w. li in images obtained on january 29 and 30 , 2002 ( nakano et al .
2002 ; ut dates are used throughout this _ letter _ ) .
the relative proximity of this event ( @xmath4 km s@xmath5 , @xmath6 mpc , tully 1988 ) and its ensuing brightness made it a promising target for intensive follow - up studies .
three independent teams promptly obtained low resolution spectroscopic observations on jan .
30 ( meikle et al . 2002 ) , and jan . 31 ( kinugasa et al . 2002 , gal - yam & shemmer 2002 ) .
all groups noted the spectral resemblance of this object to the so - called `` hypernovae '' 1998bw and 1997ef , both of which may have been associated with gamma - ray bursts ( grbs ; galama et al .
1998 , iwamoto et al .
2000 ) . using the well - sampled spectral observations of sn 1998bw by patat et al .
( 2001 ) , gal - yam & shemmer estimated that sn 2002ap was discovered prior to peak brightness .
being both nearby , of a rare type , and possibly connected to grbs , sn 2002ap became a focus of a multi - wavelength observational effort .
follow up studies include detection in the radio ( berger , kulkarni , & frail 2002 ) , ir imaging ( e.g. mattila & meikle 2002 ) and spectroscopy ( e.g. , motohara et al .
2002 ) , uv and x - ray imaging ( rodriguez - pascual et al . 2002 ) , optical polarimetry ( e.g. , wang et al . 2002 ) and high - resolution spectroscopy ( e.g. , lauroesch et al . 2002 ) .
we have undertaken an intensive observational monitoring campaign of this sn with multi - color imaging and low - resolution optical spectroscopy .
the backbone of our program consists of target - of - opportunity observations obtained with the wise observatory 1-m telescope as part of the queue program .
the resulting set of observations presents a coherent picture of the evolution of sn 2002ap in the optical band during the first month since its discovery . as this sn
now approaches solar conjunction and will not be observable during the next few months , we have analysed the data set collected so far . in this _
letter _ we briefly report and discuss our main results .
all the data presented here are available electronically via our website .
@xmath0 photometry was obtained with the wise observatory 1-m telescope during nine nights , using the tektronics ( @xmath7 pixel ) and site ( @xmath8 pixel ) ccd cameras .
time series of 60 s @xmath9band images were obtained on five occasions , in order to search for variability on short ( @xmath10 hr ) time scales . as no such effects were evident ,
these data were averaged and used to improve the photometric sampling .
figure 1 shows a combined @xmath11 image of the field of sn 2002ap , produced from the stacked images obtained during february 2002 . as the sn is isolated and contamination by underlying galactic light is constrained to lie below @xmath12 mag ( smartt , ramirez - ruiz , & vreeswijk 2002 ) , we used aperture photometry on the sn and nearby reference stars ( see fig
. 1 and table 1 ) . the sn flux was measured and compared with that of nearby bright stars using standard irafroutines .
the photometric stability of each reference star in each band was checked against the entire reference set , and stars with the lowest root - mean - square scatter values were used in the final calibration .
@xmath13 error estimates , including poisson noise in the sn flux and the scatter in the flux of reference stars , were computed , and found to be below 0.025 mag , except for the @xmath14band data , where some points have errors as large as 0.16 magnitudes . in figure 2
we plot the resulting @xmath0 light curves of this object .
the data provided by henden ( 2002 ) were used for absolute calibration .
we estimate the errors in the absolute zero point determination to be smaller than @xmath15 . as a consistency check
, we compare our photometry with the measurements of riffeser , goessl , & ries , ( 2002 ) and find excellent agreement .
we therefore combine their early @xmath16 points with our data set . as both our own observations and those of sn 1998bw around peak brightness ( galama et al .
1998 ) suggest that these objects have smoothly varying light curves , we use cubic spline interpolation to get approximate @xmath0 curves . from these , we determine the dates of peak brightness , and the sn magnitudes at peak in each band .
conservative margins on the peak dates are set by the latest point seen to be rising and the first to be declining .
assuming that the light curves of sn 2002ap had similar shapes around peak to those of sn 1998bw , we can better constrain the dates of peak by checking by how much one can adjust the peak dates without creating pronounced `` wiggles '' or `` humps '' , not seen in the light curves of galama et al . ( 1998 ) or in well sampled parts of our own data . the calculated peak dates and their estimated errors , as well as the more conservative margins discussed above ,
are reported in table 2 , along with the sn absolute magnitudes , calculated assuming the distance modulus to m74 given by tully ( 1988 ) , @xmath17 for @xmath18 km s@xmath5 mpc@xmath5 .
we correct for foreground galactic extinction according to schlegel , finkbeiner , & davis ( 1998 ) , using the extinction curve of cardelli , clayton , & mathis ( 1989 ) . based on high resolution spectroscopic observations , klose , guenther & woitas ( 2002 ) estimate that this sn suffered little intrinsic extinction ( @xmath19 mag ) in its host , so
the absolute magnitudes we derive are probably a fair estimate of the sn luminosity .
if so , it is clearly evident that unlike sn 1998bw , sn 2002ap was not optically luminous , being @xmath20 magnitudes fainter in the @xmath1band .
the peak luminosity of sn 2002ap may be comparable to that of sn 1998bw only if the distance to m74 is significantly underestimated . apart from the lower peak magnitudes , the general shape of the light curves is similar to those of sn 1998bw ( galama et al . 1998 ) . as can be seen in fig . 2 and table 2 , the light curves are wider in the redder bands , and the bluer the band , the earlier the date of the peak .
the decline rates are also similar , with the sn becoming fainter by @xmath21 mag in the @xmath9band some 20 days after peak .
.local calibrators in the field of sn 2002ap [ cols="<,<,<,^,^,^,^,^",options="header " , ] note : + the given peak dates were estimated from cubic spline interpolations and their errors estimated assuming that the light curves of sn 2002ap were similar to those of sn 1998bw around peak ( see text ) .
the most conservative range of possible dates for the peak in each band was set to lie between the latest point seen to be rising and the earliest seen to be on the decline .
+ long - slit spectra of sn 2002ap were obtained with the wise observatory 1 m telescope on five nights : 2002 january 31 and february 7 , 14 , 21 and 27 , using the faint object spectrograph and camera ( fosc ; brosch & goldberg 1994 ) .
we used a 10@xmath22wide slit , and a 600 line mm@xmath5 grism , resulting in a dispersion of 3.75 / pix ( @xmath238 resolution ) and a spectral range of @xmath234000
two exposures were obtained each night and used to reject cosmic ray hits .
reduction of the bias - subtracted and flat - field corrected spectra was carried out in the usual manner using iraf .
a he - ar arc spectrum was used for the wavelength solution . in all but the first spectrum ,
the sn and a nearby bright star were observed simultaneously through the slit ( see fig .
1 ) , providing an intrinsic relative calibration during non - photometric conditions ( see , e.g. , kaspi et al .
our photometry shows this star to be constant ( to within photometric errors ) during the period of the observations .
the wide ( @xmath24 ) slit ensured that the effects of atmospheric refraction at the non - parallactic angle were not important .
each spectrum of the sn was divided by the simultaneously - observed spectrum of the comparison star . a spectrum of the comparison star obtained on feb . 7 ,
under photometric conditions , was flux calibrated using the wise observatory standard sensitivity function and extinction curve .
these do not change appreciably from night to night , and are routinely updated using spectrophotometric standard stars .
the sn spectra were calibrated to an absolute flux scale by multiplying each sn / star ratio by the flux calibrated spectrum of the comparison star .
the january 31 spectrum ( taken without the comparison star in the slit ) was directly calibrated using the mean curves .
the final fluxed spectra were then compared to our photometry ( fig .
2 ) , showing excellent consistency .
we estimate that the relative flux calibration of the last four spectra is good to @xmath25 per cent , and the absolute flux calibration has an uncertainty of @xmath2310 per cent , at most .
figure 3 shows our entire spectroscopic data set .
the first spectrum is very blue and almost featureless , peaking around @xmath26 with secondary broad emission peaks around @xmath27 and @xmath28 .
this spectral shape closely resembles early time spectra of sn 1997ef ( e.g. garnavich et al .
1997 ) and 1998bw ( e.g. , patat et al .
2001 , see also fig .
4 , top panel ) leading to the prompt spectral identification of this object ( @xmath29 ) .
figure 4 demonstrates that this spectral resemblance continues throughout the epoch of peak brightness ( unless otherwise noted , from here on we refer to our best estimate @xmath1band peak date , feb .
7.1 ut , as the time of peak brightness ) , and up until the date of our last spectrum ( @xmath30 days after maximum ) .
the spectral evolution of sn 2002ap displays several characteristic trends also identified in 1997ef ( matheson et al .
2001 , iwamoto et al .
2000 ) and 1998bw ( stathakis et al .
2000 , patat et al .
2001 , and references therein ) .
these include the reddening of the continuum , a redward shift and narrowing of prominent `` emission - like '' and `` absorption - like''features , and decreasing values of the expansion velocities .
following patat et al .
( 2001 ) we have calculated expansion velocities from the si ii @xmath31 absorption , and find a decrease from @xmath32 km s@xmath5 @xmath33 days prior to peak to @xmath34 km s@xmath5 at peak , and to @xmath35 km s@xmath5 14 days later . this is similar to the results obtained for 1998bw , and consistent with the measurements of motohara et al .
( 2002 ; @xmath36 km s@xmath5 around peak from
@xmath37 he i ) and filippenko & chornock ( 2002 ; @xmath38 km s@xmath5
four days after peak from @xmath39 o i ) .
the prominence of the @xmath40 emission peak relative to the one bluewards of @xmath26 ( both tentatively attributed to fe ii blends , e.g. , patat et al . 2001 and references therein ) in spectra of sn 2002ap at and after peak resembles spectra of sn 1997ef but differs somewhat from those of sn 1998bw .
a notable feature in the spectra of sn 2002ap is a narrow , unresolved absorption , consistent with he ii @xmath41 .
this feature first emerges in the spectrum taken near peak brightness and is stronger still seven days later .
it does not appear in later spectra of the object , nor in spectra of either sn 1997ef or 1998bw .
we have presented the results of our optical photometric and spectroscopic monitoring of sn 2002ap , during the first month after its discovery .
our best estimate for the date of @xmath1band maximum is feb .
7.1 , 2002 , and our derived peak magnitude of m@xmath3 is significantly lower than the one measured for the `` hypernova '' sn 1998bw
. however , except for the lower peak magnitude , the characteristics of the light curves , as well as the spectral evolution of this object , are strikingly similar to those of sne 1997ef and 1998bw .
our work shows that , at least to some degree , one may use the well studied properties of sn 1998bw as a tentative basis upon which to plan future observations of sn 2002ap , after scaling the peak magnitude . in light of its similarity
to sne 1998bw and 1997ef , one of the intriguing questions about sn 2002ap is whether it was also associated with a grb . using our most conservative estimate for the date of the @xmath1band peak , feb
@xmath42 days ( table 2 ) , and assuming that the lag between a hypothetical grb and the time of @xmath1band maximum is similar to that measured for sn 1998bw ( 14.3 days , galama et al .
1998 ) , we would expect the grb trigger to have occurred around jan .
@xmath43 days . however , if we use instead our estimated @xmath14band peak date , feb .
@xmath44 days , which is best constrained by our photometry , along with the appropriate @xmath14band time lag from galama et al .
1998 ( 13.7 days ) , the resulting grb trigger time is jan .
@xmath45 days .
hurley et al . ( 2002 ) found no candidate grb that might be associated with sn 2002ap in an intensive search of gamma - ray data from all available sources , starting jan .
our results suggest that the grb trigger may have occurred outside the time frame searched .
if the grb - peak magnitude time lag for sn 2002ap was just one day longer than the lag measured for sn 1998bw , the trigger is likely to have been missed by hurley et al .
we conclude that in order to detect , or set a secure upper limit to the fluence of a grb associated with sn 2002ap , a search similar to the one reported by hurley et al . should be extended to include data taken several days prior to january 21 .
finally , there remains the question of the classification of sn 2002ap and its look - alikes , sne 1998bw and 1997ef ( and possibly also sn 1998ey , garnavich , jha , & kirshner 1998 , and the recent sn 2002bl , filippenko , leonard , & moran 2002 ) .
various authors have used various types for these events , usually some combination of ib , ic , and `` peculiar '' . in view of the fact that these sne are spectroscopically similar to each other , but well distinguished from prototypical sne ib ( e.g. sn 1984l ) or ic ( e.g. , sn 1987 m or 1994i )
, we suggest that they be defined as a new sn sub - type , and provisionally designate them as type i d sne .
this should replace the often used nickname `` hypernovae '' , initially used to describe the unusually energetic and luminous sn 1998bw ( iwamoto et al .
this term has since become somewhat misleading as it was later applied to events that were modeled as highly energetic explosions , but were not especially luminous , such as sn 1997ef ( iwamoto et al . 2000 ) .
currently , all events that are spectroscopically similar to sne 1998bw and 1997ef , e.g. , 2002ap and 2002bl , are sometimes referred to as `` hypernovae '' for lack of a better name , even though they are not known to be very luminous or exceptionally energetic
. admittedly , these events are probably closer to type ic sne than to any other sub - type , and the physical differences between them and other type ic s are not fully understood .
however , the current sn classification scheme is purely observational and based mostly on optical spectroscopic properties ( see filippenko 1997 , for a review ) . in our opinion , the difference between the early - time spectra of the three well studied `` i d '' events and those of prototypical ib and ic sne are no less significant than the differences between other types of sne .
so much so that there was some question as to whether the first of these events , 1997ef , was in fact a sn , when judged by its earliest spectra ( garnavich et al .
this work , and the queue observing program , would not have been possible without the support of the wise observatory staff , especially p. ibbetson and j. dann .
we thank n. brosch , y .- j .
choi , s. kaspi and y. lipkin for their help and support .
special thanks go to d. maoz for many helpful discussions .
v. filippenko and f. patat are thanked for making available digital copies of their data on sne 1997ef and 1998bw , and the anonymous referee for useful comments .
this work was supported by the israel science foundation the jack adler foundation for space research , grant 63/01 - 1 .
berger e. , kulkarni s. r. , & frail d. a. , 2002 , iau circ .
7817 brosch n. , goldberg y. , 1994 , mnras , 268 , l27 cardelli , j. a. , clayton , g. c. , & mathis , j. s. 1989 , apj , 345 , 245 filippenko , a. v. 1997 , ara&a , 35 , 309 filippenko a. v. , & chornock r. , 2002 , iau circ .
7825 filippenko a. v. , leonard , d. c. , & moran , e. c. 2002 , iau circ .
7845 galama t. j. , et al .
, 1998 , nature , 395 , 670 gal - yam a. , & shemmer o. , 2002 , iau circ .
7811 garnavich , p. , jha , s. , kirshner , r. , & challis , p. 1997
, iau circ .
6778 garnavich , p. , jha , s. , & kirshner , r. 1998 , iau circ .
7066 henden , a. 2002 , gcn 1242 hurley , k. et al .
2002 , gcn 1252 iwamoto , k. et al .
1998 , nature , 395 , 672 iwamoto , k. et al .
2000 , apj , 534 , 660 kaspi s. , smith .
p.s . , netzer h. , maoz d. , jannuzi b.t . , giveon u. , 2000 , apj , 533 , 631 kinugasa k. , kawakita h. , ayani k. , kawabata t. , & yamaoka h. , 2002 , iau circ .
7811 klose , s. , guenther , e. , & woitas , j. 2002 , gcn 1248 lauroesch j. t. , york d. g. , hastings c. , bowen d. v. , meyer d. , lamb d. q. , 2002 , gcn 1247 matheson , t. , filippenko , a. v. , li , w. , leonard , d. c. , & shields , j. c. 2001 , aj , 121 , 1648 mattila s. , & meikle p. , 2002
, iau circ .
7820 meikle p. , lucy l. , smartt s. , leibundgut b. , lundqvist p. , & ostensen r. , 2002 , iau circ .
7811 monet , d. , et al .
1996 , usno - sa2.0 , ( u.s . naval observatory , washington dc ) motohara k. , nomoto k. , gerardy c. , fesen r. , & henry p. , 2002
, iau circ .
7834 nakano s. , kushida r. , kushida y. , & li w. , 2002 , iau circ .
7810 patat , f. et al .
2001 , apj , 555 , 900 poznanski , d. , gal - yam , a. , maoz , d. , filippenko , a. v. , leonard , d. c. , & matheson , t. 2002 , pasp , submitted , astro - ph/0202198 riffeser , a. , goessl , a. c. , & ries , c. 2002 , iau circ .
7825 rodriguez pascual p. , gonzalez riestra r. , gonzalez garcia b. , santos lleo m. , guainazzi m. , schartel n. , 2002 , iau circ .
7821 schlegel , d. j. , finkbeiner , d. p. , & davies , m. 1998 , apj , 500 , 525 smartt s. , ramirez - ruiz e. , vreeswijk p. , 2002
, iau circ .
7816 smith , j. a. , et al .
2002 , aj , in press , astro - ph/0201143 stathakis , r. a. et al .
2000 , mnras , 314 , 807 tully , r. b. 1988 , in `` nearby galaxies catalogue '' , cambridge university press , cambridge , uk wang l. , baade d. , fransson c. , hoeflich p. , lundqvist p. , & wheeler j. c. , 2002 , iau circ . 7820 | supernova ( sn ) 2002ap in m74 was discovered on january 29 , 2002 .
being one of the nearest ( 10 mpc ) sn events in the last decades , and spectroscopically similar to the so - called `` hypernovae '' 1997ef and 1998bw , both possibly associated with gamma - ray bursts ( grbs ) , it is of great interest . shortly after its discovery , we launched an intensive photometric and spectroscopic monitoring campaign of this event , and here we report the results of the first month of observations .
we use our @xmath0 photometry to estimate the magnitudes at , and dates of , peak brightness .
our data suggest that this object reached its peak @xmath1band luminosity on feb . @xmath2
. based on its similarity to sn 1998bw , we estimate the range of possible dates for a grb that may have been associated with sn 2002ap .
we find that it may include dates outside the time frame for which all available gamma - ray data have been intensively scanned , according to recent reports .
the absolute magnitude at peak brightness of sn 2002ap ( m@xmath3 ) shows that it was significantly fainter than sn 1998bw , or normal type - ia sne , but similar to sn 1997ef .
our spectroscopic observations confirm that sn 2002ap is strikingly similar to sne 1998bw and 1997ef .
we briefly describe the spectral evolution of this object .
to assist other observers and to stimulate theoretical models , we make our entire data set publicly available in digital form .
= = = = = = = = # 1 # 1 # 1 # 1 @mathgroup@group @mathgroup@normal@groupeurmn @mathgroup@bold@groupeurbn @mathgroup@group @mathgroup@normal@groupmsamn @mathgroup@bold@groupmsamn = `` 019 = ' ' 016 = `` 040 = ' ' 336 = " 33e = = = = = = = = # 1 # 1 # 1 # 1 = = = = = = = = supernovae : individual : sn 2002ap gamma - rays : bursts . |
dolev et al . @xcite introduced the notion of _ approximate byzantine consensus _ by relaxing the requirement of _ exact _ consensus @xcite . the goal in approximate consensus is to allow the fault - free nodes to agree on values that are approximately equal to each other ( and not necessarily exactly identical ) . in presence of byzantine faults , while _ exact _ consensus is impossible in _ asynchronous _
systems @xcite , approximate consensus is achievable @xcite .
the notion of approximate consensus is of interest in _ synchronous _ systems as well , since approximate consensus can be achieved using distributed algorithms that do not require complete knowledge of the network topology @xcite .
the rest of the discussion in this paper assumes a _ synchronous _ systems .
the fault model assumed in much of the work on byzantine consensus allows up to @xmath0 byzantine faulty nodes in the network .
we will refer to this fault model as the `` @xmath0-total '' fault model @xcite . in prior work ,
other fault models have been explored as well .
for instance , in the `` @xmath0-local '' fault model , up to @xmath0 neighbors of _ each _ node in the network may be faulty @xcite , and in the @xmath0-fraction model @xcite , up to @xmath0 fraction of the neighbors of each node may be faulty . in this paper
, we consider a _
generalized _ fault model ( to be described in the next section ) .
the _ generalized fault _
model specifies a `` fault domain '' , which is a collection of feasible fault sets ( a similar fault model is recently presented in @xcite ) .
for example , in a system consisting of four nodes , namely , nodes @xmath1 and @xmath2 , the fault domain could be specified as @xmath3 .
thus , in this case , either node 1 may be faulty , or any subset of nodes in @xmath4 may be faulty . however , node 1 may not be faulty simultaneously with another node .
the new fault model is general in the sense that the other fault models studied in the literature , such as @xmath0-total , @xmath0-local and @xmath0-fraction models , are special cases of the generalized fault model .
analysis of consensus under the generalized fault model offers some new insights into how the choice of the fault model affects algorithm design .
in particular , we consider `` iterative '' algorithms for achieving approximate byzantine consensus in synchronous point - to - point networks that are modeled by arbitrary _ directed _ graphs . the _ iterative approximate byzantine consensus _ ( iabc ) algorithms of interest have the following properties , which we will soon state more formally : * _ initial state
_ of each node is equal to a real - valued _ input _ provided to that node . *
_ validity _ condition : after each iteration of an iabc algorithm , the state of each fault - free node must remain in the _ convex hull _ of the states of the fault - free nodes at the end of the _ previous _ iteration . *
_ convergence _ condition : for any @xmath5 , after a sufficiently large number of iterations , the states of the fault - free nodes are guaranteed to be within @xmath6 of each other .
this paper is a generalization of our recent work on iabc algorithms under the @xmath0-total fault model @xcite .
the contributions of this paper are as follows : * we identify a necessary condition on the communication graph for the existence of a correct iabc algorithm under the _
generalized _ fault model ( sections [ sec : iabc ] and [ s_necessity ] ) .
* we introduce a new iabc algorithm for the generalized fault model ( section [ s_algo ] ) that uses only `` local '' information . *
a transition matrix representation of the new iabc algorithm is presented ( section [ s_sufficiency ] ) .
this representation is then used to prove the correctness of the proposed algorithm ( section [ s_proof ] ) .
since the results here generalize our prior results @xcite , naturally the proof techniques used here have some similarities to the prior work .
the material in section [ s_proof ] bears the strongest similarity to our prior work .
the rest of the paper , however , presents results that provide new intuition on the problem of approximate consensus .
in particular , materials in sections [ s_necessity ] and [ s_algo ] shed light on how the fault model influences the design of iabc algorithms .
[ [ communication - model ] ] communication model : + + + + + + + + + + + + + + + + + + + + the system is assumed to be _
synchronous_. the communication network is modeled as a simple _ directed _ graph @xmath7 , where @xmath8 is the set of @xmath9 nodes , and @xmath10 is the set of directed edges between the nodes in @xmath11 .
we assume that @xmath12 , since the consensus problem for @xmath13 is trivial .
node @xmath14 can reliably transmit messages to node @xmath15 if and only if the directed edge @xmath16 is in @xmath10 .
each node can send messages to itself as well , however , for convenience , we exclude self - loops from set @xmath10 .
that is , @xmath17 for @xmath18 . with a slight abuse of terminology
, we will use the terms _ edge _ and _ link _ interchangeably in our presentation . for each node @xmath14 ,
let @xmath19 be the set of nodes from which @xmath14 has incoming edges .
that is , @xmath20 .
similarly , define @xmath21 as the set of nodes to which node @xmath14 has outgoing edges .
that is , @xmath22 .
nodes in @xmath19 and @xmath21 are , respectively , said to be incoming and outgoing neighbors of node @xmath14 . since we exclude self - loops from @xmath10 , @xmath23 and @xmath24 .
however , we note again that each node can indeed send messages to itself . [ [ generalized - byzantine - failure - model ] ] generalized byzantine failure model : + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we consider the byzantine failure model , with possible faulty nodes specified using a `` fault domain '' @xmath25 ( defined below ) .
a faulty node may _ misbehave _ arbitrarily .
possible misbehavior includes transmitting incorrect and mismatching ( or inconsistent ) messages to different neighbors .
the faulty nodes may collaborate with each other .
moreover , the faulty nodes are assumed to have a complete knowledge of the execution of the algorithm , including the states of all the nodes , the algorithm specification , and the network topology .
the generalized fault model is characterized using _ fault domain _ @xmath26 as follows : nodes in set @xmath27 may fail during an execution of the algorithm only if there exists set @xmath28 such that @xmath29 .
set @xmath27 is then said to be a _
feasible _ fault set .
set @xmath30 is said to be a fault set , if there exists @xmath31 such that @xmath29 .
thus , each set in @xmath25 specifies nodes that may all potentially fail during a single execution of the algorithm ( a similar fault model is also considered in @xcite ) .
this feature can be used to capture the notion of correlated failures .
for example , consider a system consisting of four nodes , namely , nodes 1 , 2 , 3 , and 4 .
suppose that @xmath32 this definition of @xmath25 implies that during an execution either ( i ) node 1 may fail , or ( ii ) node 2 may fail , or ( iii ) any subset of @xmath33 may fail , and no other combination of nodes may fail ( e.g. , nodes 1 and 3 can not both fail in a single execution ) . in this case
, the reason that the set @xmath34 is in the fault domain may be that the failures of nodes 3 and 4 are correlated .
the generalized fault model is also useful to capture variations in node reliability . for instance , in the above example , nodes 1 and 2 may be more reliable than nodes 3 and 4 . therefore , while simultaneous failure of nodes 3 and 4 may occur , simultaneous failure of nodes 1 and 2 is less likely . therefore , @xmath35 .
_ local knowledge of @xmath25 : _ to implement our iabc algorithm presented in section [ s_algo ] , it is sufficient for each node @xmath14 to know @xmath36 , for each feasible fault set @xmath27 . in other words , each node only needs to know the set of its incoming neighbors that may fail simultaneously .
thus , the iterative algorithm can be implemented using only `` local '' information regarding @xmath25 .
in this section , we describe the structure of the iabc algorithms of interest , and state the validity and convergence conditions that they must satisfy .
each node @xmath14 maintains state @xmath37 , with @xmath38 $ ] denoting the state of node @xmath14 at the _ end _ of the @xmath39-th iteration of the algorithm .
initial state of node @xmath14 , @xmath40 $ ] , is equal to the initial _ input _ provided to node @xmath14 . at the
_ start _ of the @xmath39-th iteration ( @xmath41 ) , the state of node @xmath14 is @xmath42 $ ] .
the iabc algorithms of interest will require each node @xmath14 to perform the following three steps in iteration @xmath39 where @xmath41 . note that the faulty nodes may deviate from this specification . 1 .
_ transmit step : _ transmit current state , namely @xmath42 $ ] , on all outgoing edges and self - loop ( to nodes in @xmath21 and node @xmath14 itself ) .
2 . _ receive step : _ receive values on all incoming edges and self - loop ( from nodes in @xmath19 and itself ) .
denote by @xmath43 $ ] the vector of values received by node @xmath14 from its incoming neighbors and itself .
the size of vector @xmath43 $ ] is @xmath44 .
3 . _ update step : _
node @xmath14 updates its state using a transition function @xmath45 as follows .
@xmath45 is a part of the specification of the algorithm , and takes the vector @xmath43 $ ] as the input . @xmath46 & = & z_i ~ ( ~r_i[t]~ ) \label{eq : z_i}\end{aligned}\ ] ] the following conditions must be satisfied by an iabc algorithm when the set of faulty nodes ( in a given execution ) is @xmath27 : * _ validity : _
@xmath47 , and all fault - free nodes @xmath48 , + @xmath38 \geq \min_{j\in{\mathcal{v}}-f}~ v_j[t-1]$ ] @xmath38 \leq \max_{j\in{\mathcal{v}}-f}~ v_j[t-1]$ ] . and
@xmath49 , @xmath50 contains elements that are in @xmath51 but not in @xmath49 .
that is , @xmath52 . ]
* _ convergence : _ for all _ fault - free _
nodes @xmath53 , @xmath54-v_j[t ] ) = 0 $ ] an iabc algorithm is said to be _ correct _ if it satisfies the above validity and convergence conditions in the given graph @xmath55 . for a given fault domain @xmath25 for graph @xmath7 ,
the objective here is to identify the necessary and sufficient conditions for the existence of a _ correct _ iabc algorithm .
in this section , we develop a necessary condition for the existence of a correct iabc algorithm .
the necessary condition will be proved to be also sufficient in section [ s_sufficiency ] . to facilitate the statement of the necessary condition
, we first introduce the notions of `` source component '' and `` reduced graph '' using the following three definitions .
[ def : decompose ] * graph decomposition : * let @xmath56 be a directed graph .
partition graph @xmath56 into strongly connected components , @xmath57 , where @xmath58 is a non - zero integer dependent on graph @xmath56 , such that * every pair of nodes * within * the same strongly connected component has directed paths in @xmath56 to each other , and * for each pair of nodes , say @xmath14 and @xmath15 , that belong to two * different * strongly connected components , either @xmath14 does not have a directed path to @xmath15 in @xmath56 , or @xmath15 does not have a directed path to @xmath14 in @xmath56 . construct a graph @xmath59 wherein each strongly connected component @xmath60 above is represented by vertex @xmath61 , and there is an edge from vertex @xmath61 to vertex @xmath62 if and only if the nodes in @xmath60 have directed paths in @xmath56 to the nodes in @xmath63 .
@xmath59 is called the of @xmath56 .
it is known that for any directed graph @xmath56 , the corresponding decomposition graph @xmath59 is a directed _ acyclic _ graph ( dag ) @xcite .
* source component * : let @xmath56 be a directed graph , and let @xmath59 be its decomposition graph as per definition [ def : decompose ] .
strongly connected component @xmath60 of @xmath56 is said to be a _ source component _ if the corresponding vertex @xmath61 in @xmath59 is reachable from any other vertex in @xmath59 .
[ def : reduced ] * reduced graph : * for a given graph @xmath7 and a feasible fault set @xmath27 , a @xmath64 is obtained as follows : * node set is obtained as @xmath65 . * for each node @xmath66 , a feasible fault set @xmath67 is chosen , and then the edge set @xmath68 is obtained as follows : * * remove from @xmath10 all the links incident on the nodes in @xmath27 , and * * for each @xmath69 and each @xmath70 , remove link @xmath71 from @xmath10 .
+ feasible fault sets @xmath67 and @xmath72 chosen for @xmath73 may or may not be identical . note that for a given @xmath55 and a given @xmath27 , multiple reduced graphs @xmath74 may exist , depending on the choice of @xmath75 sets above . for a correct iabc algorithm to exist
, the network graph @xmath55 must satisfy the necessary condition stated in theorem [ thm : nc ] below .
[ thm : nc ] suppose that a correct iabc algorithm exists for @xmath55 .
then , any reduced graph @xmath74 , corresponding to any feasible fault set @xmath27 , must contain exactly one _ source component_. a complete proof is presented in appendix [ app_s_necessity ] .
the proof is by contradiction .
let us assume that a correct iabc algorithm exists , and for some feasible fault set @xmath27 , and feasible sets @xmath67 for each @xmath48 , the resulting reduced graph contains two source components .
let @xmath76 and @xmath77 denote the nodes in the two source components , respectively .
thus , @xmath76 and @xmath77 are disjoint and non - empty .
let @xmath78 be the remaining nodes in the reduced graph .
@xmath79 may or may not be non - empty .
assume that the nodes in @xmath27 ( if non - empty ) are all faulty , and all the nodes in @xmath76 , @xmath77 , and @xmath79 ( if non - empty ) are fault - free .
suppose that each node in @xmath76 has initial input equal to @xmath80 , each node in @xmath77 has initial input equal to @xmath81 , where @xmath82 , and each node in @xmath79 has an input in the range @xmath83 $ ] . as elaborated in appendix [ app_s_necessity ] , the faulty nodes can behave in such a manner that , in each iteration , nodes in @xmath76 and @xmath77 are forced to maintain their updated state equal to @xmath80 and @xmath81 , respectively , so as to satisfy the _
validity _ condition .
this ensures that , no matter how many iterations are performed , the _ convergence _ condition can not be satisfied .
we will prove that there exists an iabc algorithm particularly _ algorithm 1 _ below that satisfies the _ validity _ and _ convergence _ conditions provided that the graph @xmath7 satisfies the necessary condition in theorem [ thm : nc ] .
this implies that the necessary condition in theorem [ thm : nc ] is also sufficient .
_ algorithm 1 _ has the three - step structure described in section [ sec : iabc ] .
this algorithm is a generalization to accommodate the generalized fault model of iterative algorithms that were analyzed in prior work @xcite , including in our own prior work as well
the key difference from previous algorithms is in the _ update _ step below . ' '' '' * algorithm 1 * ' '' '' 1 . _
transmit step : _ transmit current state @xmath42 $ ] on all outgoing edges and self - loop .
2 . receive step : receive values on all incoming edges and self - loop .
these values form vector @xmath43 $ ] of size @xmath44 ( including the value from node @xmath14 itself ) . when a fault - free node expects to receive a message from an incoming neighbor but
does not receive the message , the message value is assumed to be equal to some _
default value_. 3 .
_ update step : _ sort the values in @xmath43 $ ] in an increasing order ( breaking ties arbitrarily ) .
let @xmath84 be a vector of nodes arranged in an order `` consistent '' with @xmath43 $ ] : specifically , @xmath85 is the node that sent the smallest value in @xmath43 $ ] , @xmath86 is the node that sent the second smallest value in @xmath43 $ ] , and so on .
the size of vector @xmath84 is also @xmath44 .
+ from vector @xmath43 $ ] , eliminate the smallest @xmath87 values , and the largest @xmath88 values , where @xmath87 and @xmath88 are defined as follows : * @xmath87 is the largest number such that there exists a feasible fault set @xmath89 containing nodes @xmath90 . recall that @xmath91 .
* @xmath88 is the largest number such that there exists a feasible fault set @xmath92 containing nodes @xmath93 .
+ @xmath94 and @xmath95 above may or may not be identical .
+ let @xmath96 $ ] denote the set of nodes from whom the remaining @xmath97 values in @xmath43 $ ] were received , and let @xmath98 denote the value received from node @xmath99 $ ] .
note that @xmath100 $ ] .
hence , for convenience , define @xmath101 $ ] to be the value node @xmath14 `` receives '' from itself .
observe that if @xmath99 $ ] is fault - free , then @xmath102 $ ] .
+ define @xmath46 ~ = ~ z_i(r_i[t ] ) ~ = ~\sum_{j\in n_i^*[t ] } a_i \ , w_j \label{e_z}\end{aligned}\ ] ] where @xmath103| } = \frac{1}{|n_i^-|+1-f_1 - f_2}\ ] ] + the `` weight '' of each term on the right - hand side of ( [ e_z ] ) is @xmath104 , and these weights add to 1 .
also , @xmath105 .
although @xmath106 and @xmath104 may be different for each iteration @xmath39 , for simplicity , we do not explicitly represent this dependence on @xmath39 in the notations . ' '' '' observe @xmath107 nodes whose values are eliminated in the _ update _ step above are all in @xmath19 .
thus , the above algorithm can be implemented by node @xmath14 if it knows which of its incoming neighbors may fail simultaneously ; node @xmath14 does not need to know the entire fault domain @xmath25 as such .
the main difference between the above algorithm and iabc algorithms in prior work is in the choice of the values eliminated from vector @xmath43 $ ] in the _ update _ step .
the manner in which the values are eliminated ensures that the values received from nodes @xmath108 and @xmath109 ( i.e. , the smallest and largest values that survive in @xmath43 $ ] ) are within the convex hull of the state of fault - free nodes , even if nodes @xmath108 and @xmath109 may not be fault - free .
this property is useful in proving algorithm correctness ( as discussed below ) .
we will show that algorithm 1 satisfies validity and convergence conditions , provided that @xmath55 satisfies the condition below , which matches the necessary condition stated in theorem [ thm : nc ] .
* sufficient condition : * _ any reduced graph @xmath74 corresponding to any feasible fault set @xmath27 contains exactly one _ source component_. _ in the rest of this section , we assume that @xmath110 satisfies the above condition . to prove its sufficiency , we first develop a _ transition matrix _ representation of the _ update _ step in algorithm 1 . in our discussion
below , @xmath111 $ ] is a square matrix , @xmath112 $ ] is the @xmath14-th row of the matrix , and @xmath113 $ ] is the element at the intersection of the @xmath14-th row and @xmath15-th column of @xmath111 $ ] . for a given execution of algorithm 1 , let @xmath27 denote the actual set of faulty nodes in that execution .
let @xmath114 .
without loss of generality , suppose that nodes @xmath115 through @xmath116 are fault - free , and if @xmath117 , nodes @xmath118 through @xmath9 are faulty .
denote by @xmath119 $ ] the column vector consisting of the initial states of all the fault - free nodes .
denote by @xmath120 $ ] , where @xmath121 , the column vector consisting of the states of all the fault - free nodes at the end of the @xmath39-th iteration .
the @xmath14-th element of vector @xmath120 $ ] is state @xmath38 $ ] .
the size of vector @xmath120 $ ] is @xmath122 .
we will show that the iterative update of the state of a fault - free node @xmath123 performed in ( [ e_z ] ) in algorithm 1 can be expressed using the matrix form below .
@xmath124 = { \textbf{m}}_i[t]~v[t-1]\ ] ] where @xmath112 $ ] is a _
stochastic row _ vector of size @xmath125 .
that is , @xmath113 \geq 0 $ ] , for @xmath126 , and @xmath127 = 1 $ ] .
, the row vector @xmath112 $ ] may depend on the state vector @xmath128 $ ] as well as the behavior of the faulty nodes in @xmath27 . for simplicity , the notation @xmath112 $ ] does not explicitly represent this dependence .
] by `` stacking '' ( [ matrix : e_z ] ) for different @xmath14 , @xmath129 , we will represent the _ update _ step of algorithm 1 at all the fault - free nodes together using ( [ matrix : alg1 ] ) below .
@xmath130 = { \textbf{m}}[t]~v[t-1]\ ] ] where @xmath111 $ ] is a @xmath131 _ row stochastic _ matrix , with its @xmath14-th row being equal to @xmath112 $ ] in ( [ matrix : e_z ] ) . @xmath111 $ ]
is said to be a . in the rest of this section
, we will first `` construct '' a transition matrix @xmath111 $ ] that satisfies certain desirable properties .
then , we will identify a connection between the transition matrix and the sufficiency condition stated above , and use this connection to establish _ convergence _
property for algorithm 1 .
validity _ property also follows from the transition matrix representation .
we will construct a transition matrix with the property described in lemma [ lemma : tm2 cm ] below .
[ lemma : tm2 cm ] the _ update step _ of algorithm 1 at the fault - free nodes can be expressed using row stochastic transition matrix @xmath111 $ ] , such that there exists a feasible fault set @xmath67 for each @xmath132 such that , for all @xmath133 , @xmath134 ~ \geq ~ \beta$}}\ ] ] where @xmath135 is a constant ( to be defined later ) , and @xmath136 . in @xcite as well
, we construct a transition matrix to prove correctness of an iabc algorithm under the @xmath0-total fault model .
however , the _ generalized _ fault model introduces additional complexity , which is handled here using a new approach to construct the transition matrix .
we prove the correctness of lemma [ lemma : tm2 cm ] by constructing @xmath112 $ ] for @xmath137 that satisfies the conditions in lemma [ lemma : tm2 cm ] .
recall that @xmath27 is the set of faulty nodes , and @xmath114 . as stated before , without loss of generality , nodes @xmath115 through @xmath138 are assumed to be fault - free , and the remaining @xmath139 nodes faulty .
consider a fault - free node @xmath14 performing the _ update _ step in algorithm 1 . in the _ update _
step , recall that the smallest @xmath87 and the largest @xmath88 values are eliminated from @xmath43 $ ] , where the choice of @xmath87 and @xmath88 is described in algorithm 1 .
let us denote by @xmath140 and @xmath141 , respectively , the set of nodes and @xmath141 may be different for each @xmath39 , for simplicity , we do not explicitly represent this dependence on @xmath39 in the notations @xmath140 and @xmath141 . ] from whom the smallest @xmath87 and the largest @xmath88 values
were received by node @xmath14 in iteration @xmath39 .
define sets @xmath142 and @xmath143 to be subsets of @xmath140 and @xmath141 that contain all the fault - free nodes in @xmath140 and @xmath141 , respectively .
that is , @xmath144 and @xmath145 .
construction of @xmath112 $ ] differs somewhat depending on whether sets @xmath146 and @xmath96 \cap f$ ] are empty or non - empty .
we divide the possibilities into 6 separate cases . due to space limitation
, here we present the construction for one of the cases ( named case i ) .
the construction for the remaining cases is presented in appendix [ app_s_construction ] .
@xmath157 + \frac{l_k}{|{\mathcal{l}}_g| } \sum_{j \in { \mathcal{l}}_g } v_j[t-1 ] \label{eq : casei}\end{aligned}\ ] ] clearly , at least one of @xmath158 and @xmath159 must be @xmath160 .
we now define elements @xmath113 $ ] of row @xmath112 $ ] : * for @xmath161\cap({\mathcal{v}}-f)$ ] : in this case , @xmath15 is either a fault - free incoming neighbor of @xmath14 , or @xmath14 itself . for each such @xmath15 , define @xmath113 = a_i$ ] .
this is obtained by observing in ( [ e_z ] ) that the contribution of such a node @xmath15 to the new state @xmath38 $ ] is @xmath162 $ ] .
+ the elements of @xmath112 $ ] defined here add up to @xmath163 \cap ( { \mathcal{v}}- f)|~a_i\ ] ] * for @xmath164 : in this case , @xmath15 is a fault - free node in @xmath140 or @xmath141 .
+ for each @xmath165 , @xmath166 ~=~ a_i
\ , \sum_{k \in n_i^*[t ] \cap f } \frac{s_k}{|{\mathcal{s}}_g|}\ ] ] and for each node @xmath167 , @xmath166 ~=~ a_i \ , \sum_{k \in n_i^*[t ] \cap f } \frac{l_k}{|{\mathcal{l}}_g|}\ ] ] to obtain these two expressions , we represent value @xmath168 sent by each faulty node @xmath169 in @xmath96 $ ] , i.e. , @xmath170 \cap f$ ] , using ( [ eq : casei ] ) .
recall that this node @xmath169 contributes @xmath171 to ( [ e_z ] ) .
the above two expressions are then obtained by summing ( [ eq : casei ] ) over all the faulty nodes in @xmath96\cap f$ ] , and replacing this sum by equivalent contributions by nodes in @xmath142 and @xmath143 .
+ the elements of @xmath112 $ ] defined here add up to @xmath172 \cap f } ( s_k + l_k ) = |n_i^*[t ] \cap f|~a_i.\ ] ] * for @xmath173 \cup { \mathcal{s}}_g \cup { \mathcal{l}}_g)$ ] : these fault - free nodes have not yet been considered above . for each such node @xmath15 , define @xmath113 = 0 $ ] . with the above definition of @xmath112 $ ]
, it should be easy to see that @xmath112\,v[t-1]$ ] is , in fact , identical to @xmath38 $ ] obtained using ( [ e_z ] ) .
thus , the above construction of @xmath112 $ ] results in the contribution of the faulty nodes in @xmath96 $ ] to ( [ e_z ] ) being replaced by an equivalent contribution from fault - free nodes in @xmath143 and @xmath142 .
[ [ properties - of - textbfm_it ] ] properties of @xmath112 $ ] : + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + first , we show that @xmath111 $ ] is row stochastic .
observe that all the elements of @xmath112 $ ] are non - negative .
also , all the elements of @xmath112 $ ] above add up to @xmath163 \cap ( { \mathcal{v}}- f)|~a_i + |n_i^*[t ] \cap f|~a_i = |n_i^*[t]|~a_i = 1\ ] ] because @xmath174|$ ] as defined in algorithm 1 .
thus , @xmath112 $ ] is a stochastic row vector . recall that from the above discussion , for @xmath170 $ ] , one of @xmath158 and @xmath159 must be @xmath160 .
without loss of generality , assume that @xmath175 for some @xmath176 \cap f$ ] .
consequently , for each node @xmath165 , @xmath113 \geq \frac{a_i}{|{\mathcal{s}}_g| } s_s \geq \frac{a_i}{2|{\mathcal{s}}_g|}$ ] . also , for each fault - free node @xmath15 in @xmath96 $ ] , @xmath113 = a_i$ ] .
thus , if @xmath135 is chosen such that @xmath177 and @xmath67 is defined to be equal to @xmath141 , then the condition in the lemma holds for node @xmath14 .
that is , @xmath113 \geq \beta$ ] for @xmath133 .
[ [ all - cases - together ] ] all cases together : + + + + + + + + + + + + + + + + + + + using similar constructions in other cases as well ( presented in appendix [ app_s_construction ] ) and a suitable choice of @xmath135 ( presented in appendix [ a_s_together ] ) , we can obtain a row stochastic matrix @xmath111 $ ] , and for each @xmath132 identify a feasible fault set @xmath67 , such that @xmath113 \geq \beta$ ] for all @xmath133 .
thus , lemma [ lemma : tm2 cm ] can be proved correct .
the rest of the proof structure is derived from our previous work wherein we proved the correctness of an iabc algorithm for the @xmath0-total fault model @xcite .
let @xmath178 denote the set of all the reduced graphs of @xmath55 corresponding to a feasible fault set @xmath27 .
let @xmath179 .
@xmath180 depends on @xmath27 and the underlying network , and is finite . in this discussion , let us denote a reduced graph by an italic upper case letter , and the corresponding `` connectivity matrix '' ( defined below ) using the same letter in boldface upper case .
thus , @xmath181 denotes the connectivity matrix for graph @xmath182 .
non - zero elements of connectivity matrix @xmath181 are defined as follows : ( i ) for @xmath183 , @xmath184 if and only if @xmath185 , and ( ii ) @xmath186 for @xmath129 .
that is , non - zero elements of row @xmath187 correspond to the incoming links at node @xmath14 , and the self - loop at node @xmath14 .
thus , the connectivity matrix for any reduced graph in @xmath178 has a non - zero diagonal .
based on the _ sufficient condition _ stated at the start of section [ s_sufficiency ] and lemma [ lemma : tm2 cm ] , we can show the following key lemmas .
the proofs are presented in appendix [ a_s_non - zero ] and [ a_s_cm ] .
[ lemma : non - zero ] for any @xmath188 has at least one non - zero column .
[ lemma : cm ] for any @xmath121 , there exists a graph @xmath182 such that @xmath189 $ ] .
[ thm : sufficiency ] suppose that @xmath55 satisfies the sufficient condition stated above .
algorithm 1 satisfies both the validity and convergence conditions . a complete proof is presented in appendix [ a_sufficiency ] . by repeated application of ( [ matrix : alg1 ] )
, we can represent the _ update _ step of algorithm 1 at the @xmath39-th iterations ( @xmath121 ) as : @xmath190 & = & \left(\,\pi_{i=1}^t { \textbf{m}}[i]\,\right)\ , v[0 ] \label{e_v_t}\end{aligned}\ ] ] where @xmath191 $ ] is constructed as described above .
when presenting matrix products , for convenience of presentation , we adopt the following convention : for @xmath192 , @xmath193 $ ] denotes the `` backward '' product @xmath194{{\bf a}}[b-1]\cdots{{\bf a}}[a]$ ] . thus , @xmath195 $ ] in ( [ e_v_t ] ) above represents @xmath196{{\bf m}}[t-1]\cdots{{\bf m}}[1]$ ] .
since @xmath191 $ ] is row stochastic , then from ( [ matrix : alg1 ] ) , it follows that algorithm 1 satisfies the validity condition .
based on lemmas [ lemma : non - zero ] and [ lemma : cm ] , we can also show that the rows of @xmath197 $ ] become identical in the limit ( as elaborated in appendix [ a_sufficiency ] ) .
this observation and ( [ e_v_t ] ) together imply that the states of the fault - free nodes satisfy the convergence condition too .
this paper considers a _ generalized _ fault model , which can be used to specify more complex failure patterns , such as correlated failures or non - uniform node reliabilities . under this fault model , we prove a _ tight _ necessary and sufficient condition for the existence of synchronous iterative approximate byzantine consensus algorithms in arbitrary directed graphs .
the analysis of consensus under the generalized fault model sheds new light on how the fault model affects algorithm design . 10 d. p. bertsekas and j. n. tsitsiklis . .
optimization and neural computation series .
athena scientific , 1997 .
v. bhandari and n. h. vaidya . on reliable broadcast in a radio network .
in _ proceedings of the twenty - fourth annual acm symposium on principles of distributed computing _ , podc 05 , pages 138147 , new york , ny , usa , 2005 .
s. dasgupta , c. papadimitriou , and u. vazirani . .
mcgraw - hill higher education , 2006 .
d. dolev , n. a. lynch , s. s. pinter , e. w. stark , and w. e. weihl .
reaching approximate agreement in the presence of faults .
, 33:499516 , may 1986 .
m. j. fischer , n. a. lynch , and m. s. paterson . impossibility of distributed consensus with one faulty process . , 32:374382 , april 1985 .
j. hajnal .
weak ergodicity in non - homogeneous markov chains . in _ proceedings of the cambridge philosophical society _ , volume 54 , pages 233246 , 1958 .
r. m. kieckhafer and m. h. azadmanesh .
low cost approximate agreement in partially connected networks .
, 3(1):5385 , 1993 .
broadcast in radio networks tolerating byzantine adversarial behavior . in _ proceedings of the twenty - third annual acm symposium on principles of distributed computing _ , podc 04 , pages 275282 , new york , ny , usa , 2004 .
p. kuznetsov .
understanding non - uniform failure models .
, 106:5377 , 2012 .
l. lamport , r. shostak , and m. pease .
the byzantine generals problem . , 1982 .
h. leblanc , h. zhang , s. sundaram , and x. koutsoukos .
consensus of multi - agent networks in the presence of adversaries using only local information . , 2012
. n. a. lynch . .
morgan kaufmann , 1996 .
n. h. vaidya .
matrix representation of iterative approximate byzantine consensus in directed graphs .
n. h. vaidya , l. tseng , and g. liang .
iterative approximate byzantine consensus in arbitrary directed graphs .
volume abs/1201.4183 , 2012 .
j. wolfowitz .
products of indecomposable , aperiodic , stochastic matrices . in _ proceedings of the american mathematical society _ , volume 14 , pages 733737 , 1963 .
h. zhang and s. sundaram .
robustness of information diffusion algorithms to locally bounded adversaries .
, abs/1110.3843 , 2011 .
* appendix *
now , we present the proof for theorem [ thm : nc ] .
the proof is by contradiction .
let us assume that a correct iabc algorithm exists , and for some feasible fault set @xmath27 , and feasible sets @xmath67 for each @xmath48 , the resulting reduced graph contains two source components .
let @xmath76 and @xmath77 denote the nodes in the two source components , respectively .
thus , @xmath76 and @xmath77 are disjoint and non - empty .
let @xmath78 be the remaining nodes in the reduced graph .
@xmath79 may or may not be non - empty .
let us now assume that the nodes in @xmath27 ( if non - empty ) are all faulty , and all the nodes in @xmath76 , @xmath77 , and @xmath79 ( if non - empty ) are fault - free .
consider the case when ( i ) each node in @xmath76 has initial input @xmath80 , ( ii ) each node in @xmath77 has initial input @xmath81 , such that @xmath198 , and ( iii ) each node in @xmath79 ( if non - empty ) has an input in the interval @xmath199 $ ] . in the _ transmit step _ of iteration 1 of the iabc algorithm ,
suppose that the faulty nodes in @xmath27 ( if non - empty ) send @xmath200 on outgoing links to nodes in @xmath76 , send @xmath201 on outgoing links to nodes in @xmath77 , and send some arbitrary value in interval @xmath199 $ ] on outgoing links to nodes in @xmath79 ( if non - empty ) .
this behavior is possible since nodes in @xmath27 are byzantine faulty .
note that @xmath202 .
each fault - free node @xmath203 sends to nodes in @xmath204 value @xmath205 $ ] in iteration 1 .
consider any node @xmath206 .
since @xmath76 is a source component in the reduced graph , it must be true that @xmath207 . from nodes in @xmath208 .
thus , any incoming links in @xmath10 from the nodes in @xmath208 must have been removed when constructing @xmath68 for the reduced graph .
recall that when constructing @xmath68 , incoming links from nodes in @xmath209 are removed .
it should be noted that the algorithm is performed using the links in @xmath10 , not the reduced graph .
thus , in the _ transmit step _ , all links in @xmath10 are used . ] now , node @xmath14 receives @xmath210 from the nodes in @xmath211 , and values in @xmath83 $ ] from the nodes in @xmath208 , and @xmath80 from the nodes in @xmath212 .
figure [ f_necessity ] illustrates the behavior of faulty nodes in @xmath27 and the value received by node @xmath14 . and
the value received at node @xmath14.,scaledwidth=70.0% ] consider the following two cases : * * @xmath211 and @xmath208 are both non - empty : * in this case , @xmath213 and @xmath214 .
from node @xmath14 s perspective , consider two possible scenarios : ( a ) nodes in @xmath211 are all faulty , and the other nodes are fault - free , and ( b ) nodes in @xmath215 are all faulty , and the other nodes are fault - free . note that , since @xmath67 is a feasible fault set , @xmath216 is also a feasible fault set .
similarly , since @xmath27 is a feasible fault set , @xmath211 is also a feasible fault set .
+ in scenario ( a ) , from node @xmath14 s perspective , the fault - free nodes have sent values in interval @xmath199 $ ] , whereas the faulty incoming neighbors , i.e. , nodes in @xmath211 , have sent value @xmath210 . according to the validity condition , @xmath217 \geq m$ ] . on the other hand , in scenario ( b ) ,
the fault - free incoming neighbors have sent values @xmath210 and @xmath80 , where @xmath200 ; so @xmath217 \leq m$ ] , according to the validity condition .
since node @xmath14 does not know whether the correct scenario is ( a ) or ( b ) , it must update its state to satisfy the validity condition in both cases .
thus , it follows that @xmath217 = m$ ] . * * at most one of @xmath211 and @xmath208 is non - empty : * recall that @xmath36 and @xmath215 are both feasible fault sets . since at least one of these two sets is empty , their union , i.e. , @xmath218 , is also a feasible fault set .
+ then , from node @xmath14 s perspective , it is possible that all the nodes in @xmath218 are faulty , and the rest of the nodes are fault - free . in this situation ,
the values sent to node @xmath14 by the fault - free nodes ( which are all in @xmath212 ) are all @xmath80 , and therefore , @xmath217 $ ] must be set to @xmath80 as per the validity condition .
hence , @xmath217 = m$ ] for each node @xmath206 .
similarly , we can show that @xmath219 = m$ ] for each node @xmath220 .
now consider the nodes in set @xmath79 ( if non - empty ) .
all the values received by the nodes in @xmath79 are in @xmath199 $ ] , therefore , their new state must also remain in @xmath83 $ ] , as per the validity condition .
the above discussion implies that , at the end of iteration 1 , the following conditions hold true : ( i ) state of each node in @xmath76 is @xmath80 , ( ii ) state of each node in @xmath77 is @xmath81 , and ( iii ) state of each node in @xmath79 ( if non - empty ) is in the interval @xmath199 $ ] .
these conditions are identical to the initial conditions listed previously .
then , by a repeated application of the above argument ( proof by induction ) , it follows that for any @xmath221 = m$ ] for all nodes @xmath206 , @xmath222 = m$ ] for all nodes @xmath220 and @xmath223 \in [ m , m]$ ] for all nodes @xmath224 . since @xmath76 and @xmath77 both contain fault - free nodes , and @xmath225 , the _ convergence _ requirement is not satisfied .
this is a contradiction to the assumption that a correct iterative algorithm exists in @xmath55 .
when discussing case i in section [ s_construction ] , we deferred discussion of the other cases .
we present the construction for the rest of the cases here .
there are six cases in total : * case i : @xmath147 , and @xmath96 \cap f \neq \phi$ ] . * case ii : @xmath147 , and @xmath96 \cap f = \phi$ ] .
* case iii : @xmath226 , and @xmath96 \cap f \neq \phi$ ] . * case iv
: @xmath227 , and @xmath96 \cap f \neq \phi$ ] . * case v : @xmath228 , and @xmath96 \cap f \neq \phi$ ] .
* case vi : at most one of @xmath142 and @xmath143 is non - empty , and @xmath96 \cap f = \phi$ ] .
note that the choice of @xmath87 and @xmath88 in algorithm 1 ensures that the value from node @xmath14 itself is never dropped from @xmath43 $ ] ; therefore , @xmath100 $ ] , and @xmath96 $ ] is always non - empty .
now , we consider the case when @xmath147 , and @xmath96 \cap f = \phi$ ] . that is , when each of @xmath140 and @xmath141 contains at least one fault - free node , and @xmath96 $ ] contains only fault - free node(s ) .
in fact , the analysis of case ii is very similar to the analysis presented in section [ s_construction ] for case i when @xmath96 $ ] does contain a faulty node .
we now discuss how the analysis of case i can be applied to case ii . rewrite ( [ e_z ] ) as follows : @xmath46 & = & \frac{a_i}{2 } v_i[t-1 ] + \frac{a_i}{2 } v_i[t-1 ] + \sum_{j\in n_i^*[t ] - \{i\ } } a_iw_j \\ & = & a_iw_z + a_i w_i + \sum_{j\in n_i^*[t ] - \{i\ } } a_iw_j \end{aligned}\ ] ] in the above equation , @xmath229 is to be viewed as a `` virtual '' incoming neighbor of node @xmath14 , which has sent value @xmath230}{2}$ ] to node @xmath14 in iteration @xmath39 . with the above rewriting of state update
, the value received by node @xmath14 from itself should be viewed as @xmath231}{2}$ ] instead of @xmath42 $ ] . with this transformation , case
ii now becomes identical to case i , with virtual node @xmath229 being treated as an incoming neighbor of node @xmath14 .
in essence , a part of node @xmath14 s contribution ( half , to be precise ) is now replaced by equivalent contribution by nodes in @xmath143 and @xmath142 .
we now define elements @xmath113 $ ] of row @xmath112 $ ] : * for @xmath232 : @xmath113 = \frac{a_i}{2}$ ] .
this is obtained by observing in ( [ e_z ] ) that node @xmath14 s contribution to the new state @xmath38 $ ] is @xmath233}{2}$ ] . * for @xmath161 - \{i\}$ ] : in this case , @xmath15 is a fault - free incoming neighbor of @xmath14 . for each such @xmath15 , define @xmath113 = a_i$ ] .
this is obtained by observing in ( [ e_z ] ) that the contribution of node j to the new state @xmath38 $ ] is @xmath234 $ ] . * for @xmath235 : in this case , @xmath15 is a fault - free node in @xmath140 or @xmath141 .
+ for each @xmath165 , @xmath166 ~=~ \frac{a_i}{2 } \ , \frac{s_z}{|{\mathcal{s}}_g|}\ ] ] and for each node @xmath167 ,
@xmath166 ~=~ \frac{a_i}{2 } \ , \frac{l_z}{|{\mathcal{l}}_g|}\ ] ] + where @xmath236 and @xmath237 are chosen such that @xmath238 and @xmath239}{2 } = \frac{s_z}{2 } m_{{\mathcal{s } } } + \frac{l_z}{2 } m_{{\mathcal{l}}}$ ] .
note that such @xmath236 and @xmath237 exist because by definition of @xmath142 and @xmath143 , @xmath42 \geq w_j,~\forall j \in s_g$ ] and @xmath42 \leq w_j,~\forall j \in l_g$ ] .
then the two expressions above are obtained by replacing the contribution of the virtual node @xmath229 by an equivalent contribution by the nodes in @xmath142 and @xmath143 , respectively . * for @xmath173 \cup { \mathcal{s}}_g \cup { \mathcal{l}}_g)$ ] : these fault - free nodes have not yet been considered above . for each such node @xmath15 , define @xmath113 = 0 $ ] . by argument
similar to that in section [ s_construction ] , @xmath111 $ ] is row stochastic . without loss of generality ,
suppose that @xmath240 .
then for each node @xmath165 , @xmath113 = \frac{a_i}{2|{\mathcal{s}}_g|}s_z \geq \frac{a_i}{4|{\mathcal{s}}_g|}$ ] . also , for fault - free node @xmath15 in @xmath96-\{i\}$ ] , @xmath113 = a_i$ ] , and @xmath241 = \frac{a_i}{2}$ ] .
recall that by definition , @xmath242 . hence ,
if @xmath135 is chosen such that @xmath243 and @xmath67 is defined to be equal to @xmath141 , then the condition in the lemma [ lemma : tm2 cm ] holds for node @xmath14 .
that is , @xmath113 \geq \beta$ ] for @xmath244 .
now , we describe the construction of case iii .
the construction for case iv is very similar , and thus , is omitted here . in case iii , @xmath226 , and @xmath96\cap f\neq \phi$ ] .
thus , @xmath140 does not contain any fault - free nodes ( hence @xmath142 is empty ) .
this may be due to one of the following two reasons : ( i ) the set @xmath140 is non - empty , but all the nodes in @xmath140 are faulty , or ( ii ) set @xmath140 is empty .
assume that @xmath245 is a fault - free node , and that all the nodes in @xmath140 are faulty ( i.e. , @xmath246 ) or that @xmath140 is empty ( i.e. , @xmath247 ) .
in this case , observe that node @xmath108 must be fault - free ( otherwise , @xmath87 can not be the largest value as defined in algorithm 1 ) .
now , consider any node @xmath152 $ ] .
similar to the argument in case i , we can find weights @xmath154 and @xmath155 such that @xmath248 and @xmath249 + l_k~v_{l}[t-1]\ ] ] we now define @xmath113 $ ] for all fault - free @xmath15 . * for @xmath250 - \{d(f_1 + 1)\})\cap ( { \mathcal{v}}-f)$ ] .
that is , @xmath15 is a fault - free node in @xmath96 $ ] with the exception of @xmath108 . + for each such @xmath15 , define @xmath113 = a_i$ ] .
this is obtained by observing in ( [ e_z ] ) that the contribution of node @xmath15 to the new state @xmath38 $ ] is @xmath251 $ ] .
+ the elements of @xmath112 $ ] defined here ( including the case of @xmath252 ) add up to @xmath253 \cap ( { \mathcal{v}}- f)|-1)~a_i.\ ] ] * for nodes @xmath108 and @xmath254 : define + @xmath255 = a_i + \sum_{k \in n_i^*[t ] \cap f } a_i~s_k\ ] ] + and + @xmath256 = \sum_{k \in n_i^*[t ] \cap f } a_i~l_k\ ] ] + similar to case i presented in section [ s_construction ] , these two expressions are obtained by summing up the contribution over the faulty nodes in @xmath96 $ ] , and replacing the sum by an equivalent contribution by the nodes @xmath108 and @xmath254 , respectively , according to ( [ eq : caseii ] ) .
+ the above elements of @xmath112 $ ] add up to @xmath257 \cap f}(s_k + l_k)\right ) = ( 1+|n_i^*[t ] \cap f|)~a_i.\ ] ] * for @xmath173 \cup \{l\})$ ] : these fault - free nodes have not yet been considered above . for each such @xmath15 , define @xmath258 = 0 $ ] .
similar to case i , in case iii as well , it should be easy to see that @xmath259\,v[t-1]\ ] ] is identical to @xmath38 $ ] obtained using ( [ e_z ] ) .
[ [ properties - of - textbfm_it-1 ] ] properties of @xmath112 $ ] : + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + all the elements of @xmath112 $ ] are non - negative .
the elements of @xmath112 $ ] defined in case ii add up to @xmath253 \cap ( { \mathcal{v}}- f)|-1)~a_i + ( 1+|n_i^*[t ] \cap f|)~a_i = |n_i^*[t]|~a_i = 1\ ] ] thus , @xmath112 $ ] is a stochastic row vector . in case iii , recall that for any fault - free node @xmath15 in @xmath96 $ ] ( including @xmath260 and @xmath252 ) , @xmath113 \geq a_i$ ] .
thus , if @xmath135 is chosen such that @xmath261 and @xmath67 is defined to be equal to @xmath141 , then the condition in the lemma [ lemma : tm2 cm ] holds for node @xmath14 .
consider case v , where @xmath96 \cap f \neq \phi$ ] , and @xmath262 . in this case
, it should be easy to see that @xmath96 $ ] contains at least 3 nodes .
in particular , @xmath263 must be fault - free ( otherwise , @xmath87 can not be maximum possible ) , @xmath264 must be fault - free ( otherwise , @xmath88 can not be maximum possible ) , and there is a faulty node in @xmath96 $ ] . now this case can be handled similar to case iii analyzed above . in particular , entries in
@xmath112 $ ] are defined similarly with @xmath254 being defined equal to @xmath265 . also , define @xmath266 .
hence , it is easy to see that the properties of @xmath112 $ ] are identical to case iii presented above . here
, we consider the case when at most one of @xmath140 and @xmath141 contains a fault - free node and @xmath96 \cap f = \phi$ ] . without loss of generality ,
suppose that @xmath140 contains only faulty nodes , and @xmath141 may contain a fault - free node . in this case
, define @xmath113=a_i$ ] for @xmath99 $ ] ; define @xmath267 for all other fault - free nodes @xmath15 . also , define @xmath268 .
the properties of @xmath112 $ ] thus defined are identical to case iii above .
now , let us consider cases i - vi together . from the definition of @xmath104 in algorithm 1 , observe that @xmath269 ( because @xmath270 )
let us define @xmath271 moreover , observe that @xmath272 and @xmath273 . then define @xmath135 as @xmath274 this definition satisfies constraints on @xmath135 in cases i through vi ( conditions ( [ eq : beta_casei ] ) , ( [ eq : beta - caseii ] ) and ( [ eq : beta - caseiii ] ) ) .
thus , lemma [ lemma : tm2 cm ] holds for all six cases with this choice of @xmath135 .
here , we present the proof of the first key lemma used in the sufficiency proof .
+ * lemma [ lemma : non - zero ] * _ for any @xmath188 has at least one non - zero column . _
@xmath7 satisfies the _ sufficient condition _ stated at the start of section [ s_sufficiency ] .
therefore , there exists at least one non - faulty node @xmath169 in the reduced graph @xmath275 that has directed paths to all the nodes in @xmath275 ( consisting of the edges in @xmath275 ) .
since the length of the path from @xmath169 to any other node in @xmath275 is at most @xmath276 , the @xmath169-th column of matrix @xmath277 will be non - zero .
we use the loose bound of @xmath138 to simplify the presentation . ] here , we present the proof of the second key lemma used in the sufficiency proof .
we start with two definitions : for matrices @xmath278 and @xmath279 of identical size , and a scalar @xmath280 , @xmath281 provided that @xmath282 for all @xmath283 .
we want to prove the following lemma . * lemma [ lemma : cm ] * _ for any @xmath121 , there exists a graph @xmath182 such that @xmath189 $ ] .
_ observe that the @xmath14-th row of the transition matrix @xmath111 $ ] corresponds to the state update ( in algorithm 1 ) performed at fault - free node @xmath14 .
recall from lemma [ lemma : tm2 cm ] that @xmath113 \geq \beta$ ] for @xmath133 , where @xmath67 is a feasible fault set .
let us obtain a reduced graph @xmath56 by choosing @xmath67 for each @xmath14 as defined in lemma [ lemma : tm2 cm ] . then from the definition of connectivity matrix @xmath181 , lemma [ lemma : cm ] then follows .
when presenting matrix products , for convenience of presentation , we adopt the following convention : for @xmath192 , @xmath193 $ ] denotes the `` backward '' product @xmath194{{\bf a}}[b-1]\cdots{{\bf a}}[a]$ ] . in the discussion
below , we use boldface upper case letters to denote matrices , rows of matrices , and their elements . for instance
, @xmath284 denotes a matrix , @xmath285 denotes the @xmath14-th row of matrix @xmath284 , and @xmath286 denotes the element at the intersection of the @xmath14-th row and the @xmath15-th column of matrix @xmath284 . [ d_stochastic ] a vector is said to be _ stochastic _
if all the elements of the vector are _ non - negative _ , and the elements add up to 1 .
a matrix is said to be stochastic if each row of the matrix is a stochastic vector . for a row
stochastic matrix @xmath287 , coefficients of ergodicity @xmath288 and @xmath289 are defined as follows @xcite : @xmath290 it is easy to show that @xmath291 and @xmath292 , and that the rows of @xmath287 are all identical if and only if @xmath293 .
also , @xmath294 if and only if @xmath295 .
the next result from @xcite establishes a relation between the coefficient of ergodicity @xmath296 of a product of row stochastic matrices , and the coefficients of ergodicity @xmath297 of the individual matrices defining the product .
lemma [ claim_zelta ] is proved in @xcite .
it implies that if , for all @xmath14 , @xmath301 for some @xmath280 , where @xmath302 , then @xmath303 will approach zero as @xmath298 approaches @xmath304 .
we now define a _ scrambling _ matrix @xcite .
[ l_lambda_bound ] if any column of a row stochastic matrix @xmath284 contains only non - zero elements that are all lower bounded by some constant @xmath280 , where @xmath302 , then @xmath284 is a scrambling matrix , and @xmath307 .
since the above product consists of @xmath310 connectivity matrices corresponding to graphs in @xmath312 , at least one of the connectivity matrices corresponding to the @xmath180 distinct graphs in @xmath312 , say matrix @xmath313 , will appear in the above product at least @xmath138 times .
now observe that : ( i ) by lemma [ lemma : non - zero ] , @xmath314 contains a non - zero column , say the @xmath169-th column is non - zero , and ( ii ) all the @xmath309 $ ] matrices in the product contain a non - zero diagonal . these two observations together imply that the @xmath169-th column in the above product is non - zero .
let us now define a sequence of matrices @xmath315 , @xmath316 , such that each of these matrices is a product of @xmath310 of the @xmath196 $ ] matrices . specifically ,
@xmath317 \label{e_q_i}\end{aligned}\ ] ] from ( [ e_v_t ] ) and ( [ e_q_i ] ) observe that @xmath318 & = & \left(\ , \pi_{i=1}^k ~ { { \bf q}}(i ) \,\right)~{v}[0]\end{aligned}\ ] ] @xmath315 is a product of row stochastic matrices ( @xmath196 $ ] ) ; therefore , @xmath315 is row stochastic . from lemma [ lemma :
cm ] , for each @xmath320 , @xmath321 ~ \leq ~ { { \bf m}}[t]\ ] ] therefore , @xmath322 ~
\leq ~ \pi_{t=(i-1)\tau(n-\psi)+1}^{i\tau(n-\psi ) } ~ { { \bf m}}[t ] ~ = ~ { { \bf q}}(i)\ ] ] by using @xmath323 in lemma [ l_product_h ] , we conclude that the matrix product on the left side of the above inequality contains a non - zero column
. therefore , @xmath315 on the right side of the inequality also contains a non - zero column .
observe that @xmath310 is finite , and hence , @xmath324 is non - zero . since the non - zero terms in
@xmath309 $ ] matrices are all 1 , the non - zero elements in @xmath325 $ ] must each be @xmath326 1 . therefore , there exists a non - zero column in @xmath315 with all the elements in the column being @xmath327 .
therefore , by lemma [ l_lambda_bound ] , @xmath328 , and @xmath315 is a scrambling matrix . using lemma [ claim_zelta ] and the definition of @xmath315 , and using the inequalities @xmath330)\leq 1 $ ] and @xmath331
, we get @xmath332 ) ~=~ \lim_{t\rightarrow\infty } \delta\left ( \left(\pi_{i=(\lfloor\frac{t}{\tau(n-\psi)}\rfloor)\tau(n-\psi)+1}^t { { \bf m}}[i]\right ) \left(\pi_{i=1}^{\lfloor\frac{t}{\tau(n-\psi)}\rfloor } { { \bf q}}(i)\right)\right )
\\ ~ \leq ~ \lim_{t\rightarrow\infty } \pi_{i=1}^{\lfloor\frac{t}{\tau(n-\psi)}\rfloor } \lambda({{\bf q}}(i ) ) ~ = ~ 0 \end{aligned}\ ] ] thus , the rows of @xmath195 $ ] become identical in the limit . this observation , and the fact that @xmath329=(\pi_{i=1}^t { { \bf m}}[i]){v}[0]$ ] together imply that the states of the fault - free nodes satisfy the convergence condition | in this work , we consider a _
generalized _ fault model that can be used to represent a wide range of failure scenarios , including correlated failures and non - uniform node reliabilities .
this fault model is general in the sense that fault models studied in prior related work , such as @xmath0-total and @xmath0-local models , are special cases of the generalized fault model . under the generalized fault model ,
we explore iterative approximate byzantine consensus ( iabc ) algorithms in arbitrary directed networks .
we prove a necessary and sufficient condition for the existence of iabc algorithms .
the use of the generalized fault model helps to gain a better understanding of iabc algorithms . |
in 1980 culler and shalen introduced in @xcite an algebraic set in a complex space for a finitely presented group @xmath1 , now known as the character variety of @xmath1 . the framework of character varieties has been giving powerful tools and is now playing important roles in geometry and topology . on the other hand , it is not easy to calculate character varieties and thus to investigate the geometric structures in general , though an underlying idea of character varieties is simple as follows .
let @xmath1 be a finitely presented group generated by @xmath2 elements @xmath3 .
for a representation @xmath4 , the character @xmath5 of @xmath6 is the function on @xmath1 defined by @xmath7 ( @xmath8 ) . throughout this paper , we simply denote by @xmath9 the trace @xmath10 for an unspecified representation @xmath4 .
we sometimes omit the brackets in the trace like @xmath11 for simplicity . by @xcite ( see also @xcite ) , _ the @xmath12-trace identity _
@xmath13 shows that @xmath9 is expressed by a polynomial in @xmath14 , @xmath15 and @xmath16 , for any @xmath17 .
then the character variety of @xmath1 , denoted by @xmath18 , is constructed basically by the image of the set of characters @xmath19 of @xmath12-representations of @xmath1 under the map is not injective , however , for the characters of irreducible representations it is injective .
refer to @xcite . ]
@xmath20 the resulting set turns out to be a closed algebraic set ( refer to @xcite ) . by definition ,
this algebraic set depends on a choice of generators of @xmath1 ( the coordinates of @xmath18 vary if we change the choice of generating set of @xmath1 ) , however , the geometric structures do not depend on that choice .
so @xmath18 is an invariant of @xmath1 up to _ isomorphism of algebraic sets_. here two algebraic sets @xmath21 and
@xmath22 in a complex space @xmath23 are said to be isomorphic or _
bipolynomial _ if there exist polynomial maps @xmath24 and @xmath25 such that @xmath26 , @xmath27
. then @xmath28 and @xmath29 are called isomorphisms or bipolynomial maps . as far as the authors know , there exist few discussions on isomorphic transformations of character varieties . in this paper ,
we first give two polynomial maps giving isomorphisms between character varieties of a group @xmath1 coming from two different choices of generating sets , especially in the case where @xmath1 is the fundamental group of a 3-manifold associated to _ a knot _ in 3-sphere as follows .
let @xmath30 denote a 3-sphere .
for a knot @xmath31 in @xmath30 , which is an embedding of a circle @xmath32 in @xmath30 , we denote by @xmath33 the knot group of @xmath31 , i.e. , the fundamental group of the knot complement @xmath34 .
for example , there exist knots parametrized by sequences of integers @xmath35 $ ] associated to the number of _ twists _ , called _ 2-bridge knots_. ( see figure [ 2-bridge_knots ] . for more details , refer to @xcite . )
the knot @xmath36 $ ] depicted in figure [ twist_knots ] is a type of 2-bridge knots , called the _
@xmath37-twist knot_. sometimes the sequence @xmath35 $ ] is encoded in a rational number @xmath38 @xmath39 by the following continuous fraction : @xmath40 then we denote by @xmath41 the 2-bridge knot with a rational number expression @xmath38 .
it is shown that @xmath42 and @xmath43 are coprime odd integers ( refer to @xcite ) .
@xmath44= \begin{minipage}{10cm}\includegraphics[width=\hsize]{2-bridge.eps}\end{minipage}\ ] ] by _
wirtinger s algorithm _ and reduction of generators , we have the following presentation of the knot group @xmath45 : @xmath46 where @xmath47 are _ meridians _ shown in figure [ 2-bridge_knots ] , @xmath48 and @xmath49 . here
@xmath50 denotes the maximal integer @xmath37 satisfying @xmath51 . by @xcite , the character variety @xmath52 is given as the algebraic set defined by @xmath53 for example , this equation induces the following presentation of the character variety @xmath54 of the @xmath55-twist knot @xmath56 .
let @xmath57 be the chebyshev polynomial of the second kind , i.e. , for any integer @xmath2 , @xmath58 note that @xmath59 holds for any integer @xmath2 .
we denote by @xmath60 the polynomial in the complex polynomial ring @xmath61 $ ] defined by @xmath62 where @xmath63 .
using @xmath60 , we can describe the character variety @xmath64 as the algebraic set @xmath65 where @xmath66 and @xmath67 .
( see subsection [ subsec_km ] for details . for @xmath68 , refer to subsection [ subsec_k_odd ] . )
@xmath69=\begin{minipage}{7cm}\includegraphics[width=\hsize]{k - m.eps}\end{minipage}.\ ] ] on the other hand , we can also calculate the character variety using _ the kauffman bracket skein module_. ( refer to @xcite . see also subsection [ top_back ] . ) from the kauffman bracket skein theory , the following is a basic presentation of @xmath70 .
for any integer @xmath37 , let @xmath71 be the polynomial in @xmath61 $ ] defined by @xmath72 and let @xmath73 be the second factor of @xmath71 .
then the character variety @xmath74 is also described as the algebraic set @xmath75 where @xmath76 and @xmath77 ( refer to @xcite ) . by definition , the algebraic sets in and at @xmath78 must be isomorphic as algebraic sets . in subsection [ subsec_km ] , we construct bipolynomial maps which give isomorphisms between and at @xmath78 . actually , the presentation in of @xmath74 is more useful than to discuss a geometric property of @xmath18 , that is , the number of irreducible components .
for example , the presentation of @xmath74 immediately shows the following result .
[ main_thm ] for any positive integer ( @xmath79 ) , taking the mirror image of @xmath80 and arranging it , we can obtain @xmath81 and thus a similar result to theorem [ main_thm ] . in that case , @xmath82 will shift to @xmath83 . ]
@xmath37 , the polynomial @xmath84 is irreducible in @xmath61 $ ] .
therefore , @xmath74 consists of two irreducible components .
indeed , as in @xcite the factor @xmath73 @xmath85 can not be factorized as @xmath86 where @xmath87 $ ] .
moreover , @xmath88 where @xmath87 $ ] can not occur either as a factorization of @xmath89 , since @xmath90 and @xmath91 are relatively prime in @xmath92 $ ] .
hence @xmath73 is irreducible in @xmath61 $ ] and this completes the proof of theorem [ main_thm ] .
it was studied in @xcite that the character variety of most _ double twist knots _ , including twist knots , consists of two irreducible components , however , they used the genera of algebraic varieties to show the fact .
our proof of theorem [ main_thm ] shown above uses only basic calculations on the chebyshev polynomials @xmath57 .
this is an efficiency of chebyshev polynomials .
in fact , the method used in the proof of theorem [ main_thm ] can be applied to another type of 2-bridge knots denoted by @xmath93 , which was considered in @xcite .
then that gives the following theorem using basic calculations .
[ thm_bp3 ] the character variety @xmath94 consists of exactly two irreducible components .
this paper mainly shows * presentations of the character varieties of 2-bridge knots @xmath95 and @xmath93 using the chebyshev polynomials @xmath57 , including the transformations between the algebraic sets in and , * an elementary proof of theorem [ thm_bp3 ] using chebyshev polynomials .
these recover the results of burde on the irreducibility of non - abelian @xmath0-representation spaces in @xcite .
moreover , they can be applied to determine some minimal elements of a partial ordering of prime knots ( see section [ sec_appli ] ) .
for the twist knot @xmath96 , we have @xmath97 where @xmath98 . let @xmath99 , @xmath100 and @xmath101 .
we first focus on the chebyshev polynomials @xmath102 .
the following lemma is standard , see e.g. ( * ? ? ?
* lemma 2.2 ) .
suppose the sequence @xmath103 satisfies the recurrence relation @xmath104
. then @xmath105 .
[ chev ] applying lemma [ chev ] , we have @xmath106 since @xmath107 . therefore the character variety @xmath64 is given by the presentation in ( [ ln ] ) .
now we construct two polynomial maps @xmath28 and @xmath29 that give isomorphisms between the algebraic sets in and .
define a polynomial @xmath108 in @xmath61 $ ] by @xmath109 in fact , by @xcite the general term of @xmath110 is @xmath111 and then we can check that @xmath112 holds .
now let @xmath113 be the maps @xmath114 then we obtain @xmath115 we will show that @xmath116 on the algebraic set @xmath117 at @xmath78 and @xmath118 on @xmath119 . to prove these , we will need the following lemmas .
[ s0 ] for any non - negative integers @xmath120 and @xmath121 , @xmath122 lemma [ s0 ] and @xmath123 immediately show lemmas [ s ] and [ s2 ] .
@xmath124 holds for any integer @xmath37 .
[ s ] @xmath125 holds for any integer @xmath37 .
[ s2 ] one has @xmath126 [ x ] let @xmath127 .
then we have @xmath128 hence @xmath129 one has @xmath130 [ r ] we have @xmath131 by lemma [ x ] .
one has @xmath132 [ x2 ] we have @xmath133 hence @xmath132 now we can show one of the desired propositions .
note that @xmath134 .
[ gf ] the following holds for any integer @xmath37 . @xmath135
hence @xmath136 this is equivalent to lemma [ r ] which says that @xmath130 the proposition follows .
we can also show the remaining desired proposition .
the following holds for any integer @xmath2 .
@xmath137 hence @xmath138 .
[ fg ] let @xmath139 by direct calculations one can check that @xmath140 for @xmath141 . by lemma [ x2 ] , we have @xmath142 it follows that @xmath143 since @xmath144 .
to prove proposition [ fg ] , it suffices to show that @xmath145 by lemma [ s ] , we obtain @xmath146s^2_{n-1}(t)-1,\end{aligned}\ ] ] let @xmath147 . then @xmath148 . since by lemma [ s2 ] @xmath149
, it follows that @xmath150 it is easy to check that @xmath151 .
. holds true for all @xmath2 .
this completes the proof proposition [ fg ] . for the twist knot @xmath153 , @xmath154 where @xmath155 . in this case , we let @xmath156 , @xmath157 and @xmath158 . then as in the case of @xmath56 it follows from lemma [ chev ] that @xmath159 so the character variety @xmath152 is described as the algebraic set @xmath160 where @xmath161 .
now let @xmath113 be the maps @xmath162 then we have @xmath163 note that @xmath134 and thus by proposition [ gf ] the equation @xmath118 holds on the algebraic set in .
so we will show the remaining equation @xmath116 on the algebraic set in at @xmath164 .
the following holds for any integer @xmath2 .
@xmath165 hence @xmath166 .
[ odd ] the proof is similar to that of proposition [ fg ] .
let @xmath167 by direct calculation , one can check that @xmath168 for @xmath141 . since by lemma [ x2 ] @xmath169 the following holds @xmath170 to prove proposition [ odd ] , it suffices to show that @xmath171 by lemma [ s ] , we have @xmath172s_n(t)s_{n-1}(t)+1-y,\end{aligned}\ ] ] let @xmath173 . then @xmath174 . since by lemma [ s0 ] @xmath175
, it follows that @xmath176 it is not hard to check that @xmath177 .
hence eq . holds true for all @xmath2 .
proposition [ odd ] follows .
the difference between the presentations in and or and is the method to calculate the character varieties . as seen in subsections [ subsec_km ] and [ subsec_k_odd ] , the presentations and are calculated along the definition using the trace functions of representations . on the other hand ,
the presentation in is calculated by using the kauffman bracket skein algebra . ]
( kbsa for short ) .
the kbsa of a 3-manifold is the quotient of the module over @xmath178 generated by all free homotopy classes of loops in @xmath179 by _ the kauffman bracket skein relations _ : @xmath180{skein - rel.eps}\end{minipage},\ \begin{minipage}{2cm}\includegraphics[width=\hsize]{circ.eps}\end{minipage } = -2,\ ] ] where in the first relation loops coincide each other outside dashed circles ( refer to @xcite ) . by the result of @xcite or @xcite ,
a loop @xmath121 can be considered as @xmath181 and _ the kauffman bracket skein relations _ can be thought of as the @xmath12-trace identities .
this correspondence gives a method to calculate character varieties using the kauffman bracket skein theory . from the kbsa point of view
, we can take two free homotopy classes of loops represented by @xmath182 and @xmath183 as the generators of the kbsa of @xmath184 ( see ( * ? ? ?
* section 1 ) ) .
these give the parameters @xmath76 and @xmath185 of the presentation in and thus @xmath74 .
if we want to transform the algebraic set in or to , we need to express the parameters @xmath66 and @xmath186 in or as polynomials in @xmath187 and @xmath188 . the parameter @xmath189 is exactly @xmath190 . on the other hand
, @xmath191 is @xmath192 for the loop @xmath193 shown in figure [ loop_xm ] freely homotopic to @xmath194 . in the kbsa , @xmath195 is considered as the loop @xmath193 and then it follows from the kauffman bracket skein relations that @xmath196 , @xmath197 and @xmath198 .
( see @xcite .
note that the loop @xmath110 defined in @xcite is actually the mirror image of the above @xmath193 , however , the recursion relation does not change under the mirror image . )
the general term @xmath193 is the polynomial shown in substituted @xmath199 and @xmath200 .
therefore we defined @xmath108 as in and the map @xmath201 by @xmath202 , which comes from @xmath203 .
@xmath204{loops_y.eps } \end{minipage}.\ ] ] conversely , if we want to transform the algebraic set in to or , we need to express the parameters @xmath76 and @xmath205 of as polynomials in @xmath66 and @xmath186 . by the same reason , @xmath187 is exactly @xmath206 , and @xmath207 is @xmath208 .
therefore we define the map @xmath209 by @xmath210 , where @xmath211 .
this map comes from @xmath212 .
let @xmath213 and @xmath214 . in general , it follows from @xcite that the polynomial @xmath215 , whose zero set coincides with the character variety @xmath216 , is described by @xmath217 where @xmath218 is the polynomial in @xmath219 $ ] defined by @xmath220 here if @xmath221 is a word then @xmath222 denotes the word obtained from @xmath221 by deleting the two letters at the two ends .
then @xmath223 means the element obtained from @xmath224 by applying the deleting operation @xmath225 times .
the above presentation of @xmath216 is useful to show the irreducibility of @xmath218 over @xmath178 in the case of @xmath93 . for the 2-bridge knot @xmath93
, one has @xmath226 where @xmath227 .
[ m=3 ] the proof of proposition [ m=3 ] will be presented at the end of this section .
we will need the following result , which is a generalization of the method used in @xcite and the proof of theorem [ main_thm ] , in order to show the irreducibility of @xmath218 over @xmath178 in the case of @xmath93 .
suppose @xmath228 is a polynomial in @xmath219 $ ] such that @xmath229 is an odd number , and @xmath230 and @xmath231 are relatively prime in @xmath232 $ ] .
then @xmath233 is irreducible in @xmath219 $ ] .
[ irred ] assume @xmath233 is reducible in @xmath219 $ ] .
since @xmath234 , we must have @xmath235 where @xmath236 s are polynomials in @xmath232 $ ] .
is equivalent to @xmath237 it follows that @xmath238 hence @xmath239 is an even number , a contradiction .
[ thm_irr_bp3 ] for the 2-bridge knot @xmath93 , @xmath218 is irreducible in @xmath219 $ ] .
[ irreducibility ] we first show theorem [ thm_irr_bp3 ] . by proposition [ m=3 ] , @xmath240 , where @xmath241 since @xmath242 is an odd number , by lemma [ irred ] , @xmath243 $ ] is irreducible if @xmath244 . the following lemma is standard , see e.g. @xcite . for @xmath245 , one has 1 .
@xmath57 is a monic polynomial of degree @xmath2 whose @xmath2 roots are exactly @xmath246 , @xmath247 .
2 . @xmath248 is a monic polynomial of degree @xmath2 whose @xmath2 roots are exactly @xmath249 , @xmath250 .
[ roots ] @xmath251 .
[ gcd1 ] by lemma [ roots ] ( 2 ) , it suffices to show that @xmath252 where @xmath253 and @xmath254 .
it is easy to see that holds true if @xmath255 .
recall that @xmath256 and @xmath227 . since @xmath257 is equal to either @xmath258 or @xmath259 , and @xmath260 is equal to either @xmath261 or @xmath262 ( note that @xmath263 ) , @xmath257 is equal to either @xmath264 or @xmath265 .
it follows that @xmath266 is a divisor of @xmath267 .
since @xmath268 is odd , we must have @xmath269
. @xmath270 .
[ gcd2 ] the proof is similar to that of lemma [ gcd1 ] .
we now finish the proof of theorem [ irreducibility ] . from lemmas
[ roots ] , [ gcd1 ] and [ gcd2 ] , we have @xmath271 .
hence lemma [ irred ] implies that @xmath218 is irreducible in @xmath219 $ ] for the two - bridge knot @xmath93 and this completes the proof of theorem [ thm_irr_bp3 ] . in the remainder of this section
, we dedicate to the proof of proposition [ m=3 ] . for @xmath272 , let @xmath273 then @xmath274 and @xmath275 .
let @xmath276 and @xmath277 for @xmath272 , where @xmath278 . 1 . if @xmath279 , then @xmath280 2 .
if @xmath281 , then @xmath282 [ lt ] see ( * ? ? ?
* proposition a.3 ) . for the two - bridge knot @xmath93
, we have @xmath283 if @xmath284 and @xmath285 if @xmath286 , where @xmath287 . :
@xmath288 . since @xmath289 , by lemma [ lt ] , @xmath290 note that @xmath291 and @xmath292 . hence @xmath293 where @xmath294 are the chebyshev polynomials defined by @xmath295 and @xmath296 . in particular , @xmath297 : @xmath298 .
since @xmath289 , by lemma [ lt ] , @xmath299 it follows that @xmath300 : @xmath301 . since @xmath302 , by lemma [ lt ] , @xmath303
hence @xmath304 is equal to @xmath305 where @xmath306 to proceed , we need the following lemma the followings holds .
1 . @xmath307 , 2 .
@xmath308 , 3 .
@xmath309 . [ pqr ]
@xmath310 is shown by @xcite ( see also ( * ? ? ?
* proposition a.2 ) ) , @xmath311 . to show @xmath312 ,
let @xmath313 then @xmath314 . if @xmath315 is even then @xmath316 if @xmath317 is odd then @xmath318 in both case @xmath319 hence @xmath320 to show @xmath321 , let @xmath322 then @xmath323 .
note that @xmath324 .
if @xmath315 is even then @xmath325 if @xmath317 is odd then @xmath326 in both cases @xmath327 .
hence @xmath328 from lemma [ pqr ] , we get @xmath329 this completes the proof of proposition [ m=3 ] .
theorem [ main_thm ] determines some minimal elements for a partial order on the set of prime knots in @xmath30 defined as follows ( refer to @xcite for example ) .
let @xmath31 and @xmath330 be prime knots in @xmath30 which are _ non - trivial _
, i.e. , they do not bound embedded disks in @xmath30
. then we write @xmath331 if there exists an epimorphism ( a surjective group homomorphism ) from @xmath33 onto @xmath332 .
this defines a partial order on the set of prime knots .
we can apply the following theorem to the partial order @xmath333 .
[ thm - bbrw ] suppose @xmath334 is a hyperbolic knot in @xmath30 such that @xmath335 of @xmath31 has only one irreducible component that contains the characters of irreducible representations
. then @xmath33 does not surject onto the knot group of any other non - trivial knot .
we remark that corollary 1.3 in @xcite and corollary 7.1 in @xcite also give the following ( compare to theorem [ thm - bbrw ] ) .
[ thm - ors ] let @xmath31 be a hyperbolic @xmath267-bridge knot , @xmath330 a non - trivial knot .
suppose there exists an epimorphism @xmath336 .
then @xmath330 is a @xmath267-bridge knot and furthermore @xmath335 has more than one irreducible components containing the characters of irreducible representations .
since any twist knot is a 2-bridge knot , combining theorems [ main_thm ] and [ thm - bbrw ] ( or theorem @xmath337 ) , we obtain the following corollary .
[ cor_minimal ] for any integer @xmath338 , at which the @xmath37-twist knot @xmath95 is non - trivial , @xmath95 is a minimal element for the partial order @xmath333 .
note that corollary [ cor_minimal ] was shown by kitano and suzuki @xcite in the case of @xmath339 , which is the trefoil knot ( i.e. , non - hyperbolic ) and to which we can not apply theorems [ thm - bbrw ] and [ thm - ors ]
. we can also apply theorems [ thm_bp3 ] and [ thm - bbrw ] to get the following .
[ cor_bp3 ] for any odd integer @xmath340 satisfying @xmath263 , where @xmath93 is hyperbolic , @xmath93 is a minimal element for the partial order @xmath333 .
corollaries [ cor_minimal ] and [ cor_bp3 ] also show the minimality of twist knots @xmath95 ( @xmath341 ) and the 2-bridge knots @xmath93 with respect to the partial order introduced by silver and whitten @xcite ( see also @xcite ) .
the first author has been partially supported by mext kakenhi for young scientists ( b ) grant number 22740048 .
the second author would like to thank t.t.q .
le for helpful discussions .
le : _ varieties of representations and their subvarieties of cohomology jumps for knot groups _ , ( russian ) mat .
* 184 * ( 1993 ) , 57 - 82 ; translation in russian acad .
sb . math .
* 78 * ( 1994 ) , 187 - 209 .
w. neumann and a. reid : _ arithmetic of hyperbolic manifolds _ , in topology 90 , contrib .
semester low dimensional topol .
1990 , ohio state university mathematics research institute publications * 1 * ( 1992 ) , 273310 . | in this paper , we use chebyshev polynomials to give presentations of the character varieties of certain types of 2-bridge knots .
this gives us an elementary method using basic calculations to discuss the number of irreducible components of the character varieties and thus to recover the results of burde on the irreducibility of non - abelian @xmath0-representation spaces in @xcite .
these results can be applied to determine some minimal elements of a partial ordering of prime knots . |
non - equilibrium properties associated with the peak - effect ( pe ) in the c15-laves phase superconductor ceru@xmath0 have drawn some attention in recent years @xcite . it is observed that in certain field ( h ) temperature ( t ) regime ( which includes part of the pe regime ) , the field - cooled ( fc ) vortex state in ceru@xmath0 is more disordered than the vortex state obtained by isothermal variation of h after zero field cooling ( zfc ) .
this fc vortex state in ceru@xmath0 is quite metastable in nature and susceptible to external fluctuations @xcite .
the origin of pe and the associated non - equilibrium properties in ceru@xmath0 remain a matter of debate and various possibilities , starting from a underlying phase transition@xcite to dynamical crossover in flux - pinning properties @xcite , have been put forward . in a recent communication @xcite
we have drawn analogy between the zfc / fc response of the vortex state in and around the pe regime of ceru@xmath0 and the zfc / fc response of a random field ising systems ( rfim ) . through dc magnetization study in polycrystalline samples of ceru@xmath0
, we had shown that the magnetization hysteresis in the fc state is distinctly more than the zfc state .
this relatively large hysteresis in the fc state suggested that the fc state was more disordered .
it should be recalled here that in the rfim systems while there exists a long range order in the zfc state , the same is absent in the fc state ; instead the fc state is frozen into a disordered domain structure , which is quite metastable in nature @xcite .
the analogous behaviour ( if not the exact similarity ) of these two disparate class of systems has motivated us to investigate more in this direction . in this communication
we shall present results of transport properties measured on good quality single crystals of ceru@xmath0 and v@xmath1si .
while the present results on ceru@xmath0 will enhance our earlier picture @xcite obtained through dc magnetization study on polycrystalline samples , the additional results on the well known superconductor v@xmath1si will probably indicate more universality of the concerned features .
we point out here that , although there exists report of pe in v@xmath1si @xcite , our results ( described below ) will show the existence of history effect and metastability associated with the pe of v@xmath1si .
the single crystal samples of ceru@xmath0 and v@xmath1si used in this study are prepared by dr .
a. d. huxley @xcite and dr .
a. menovsky @xcite respectively , and the details of the sample preparation and characterization can be found in the references 13 and 14 .
the residual resistivity ratio for the ceru@xmath0 and v@xmath1si samples are 15 and 47 respectively @xcite .
the electrical resistance measurements are performed using the standard linear four probe method .
a superconducting magnet and cryostat system ( oxford instruments , uk ) is used to obtain the required temperature ( t ) and magnetic field ( h ) environment . in the configuration of our measurement the current ( i@xmath2 ) is passed along the @xmath3 direction for the ceru@xmath0 sample and the @xmath4 direction for the v@xmath1si sample . in all the measurements
the direction of i@xmath2 is kept perpendicular to h. the t@xmath5 for the ceru@xmath0 and v@xmath1si samples ( obtained from our zero field resistance measurements ) are 6.1 and 16.5k respectively . the magnetic field dependence of the resistance ( r(h ) )
is measured , 1 . on zero field cooling
( zfc ) the sample to various t ( @xmath6t@xmath5 ) followed by an isothermal variation of h. 2 . on field cooling ( fc ) the sample across t@xmath5 in a fixed h to t ( @xmath6t@xmath5 ) .
this is done for various h at each t.
for a sample of type - ii superconductor with pinning , the critical current ( i@xmath5 ) decreases with the increase in h and goes to zero at the irreversibility field ( h@xmath7h@xmath8 ) .
however , for superconductors showing pe , i@xmath5(h ) shows a peak or local maximum at an intermediate h value before finally going to zero at h@xmath9 . in an isothermal field variation
, the consequence of pe will show up in r(h ) if the transport current ( i@xmath2 ) used in the measurement is larger than i@xmath5(h ) for an intermediate h regime but smaller than i@xmath5(h ) in the pe regime .
in such a situation one would observe a flux flow resistivity at fields h where i@xmath10i@xmath5(h ) but the resistance will once again fall back to zero in the pe regime where i@xmath11 i@xmath5(h ) . adjusting our measuring current i@xmath2 appropriately , we show in fig .
1(a ) and 1(b ) the r vs h plots with distinct signatures of pe for ceru@xmath0 and v@xmath1si .
( although we have results obtained with various other measuring currents at different temperatures , for the sake of clarity and conciseness we shall use these two representative figures for further discussions in the present work ) .
the intermediate flux - flow regime ( 0.8t@[email protected] t for ceru@xmath0 and 1.9t@[email protected] t for v@xmath1si ) and the pe regime ( 1.25t@[email protected] t for ceru@xmath0 and 3.48t@[email protected] t for v@xmath1si ) are very clear in these figures .
as h approaches h@xmath9 , the flux - flow resistivity starts appearing again which ultimately leads to the normal state behaviour at h@xmath8 .
the pe regime for the ceru@xmath0 sample identified in our present transport measurements agrees well with that obtained earlier with the magnetic measurements @xcite .
a comparison with the sole magnetic measurement ( to our knowledge ) on v@xmath1si@xcite leads to a similar conclusion .
we shall now focus on the field - history dependence of the resistance in both ceru@xmath0 and v@xmath1si .
the value of i@xmath2 was chosen for the two samples was such that the intermediate field flux - flow was observed in both the systems when the field is varied isothermally in the zfc mode .
the measured resistance is , however , found to be zero in the same field regime , when the measurement is performed following the fc protocol ( see fig.2 ) .
this clearly indicates that in this intermediate field regime i@xmath5(h ) is greater than i@xmath2 in the fc mode while i@xmath13 is greater than i@xmath5(h ) in the zfc mode .
thus , i@xmath5(h ) is higher in the fc mode than in the zfc mode . a related history dependence of i@xmath5 in polycrystalline sample of ceru@xmath0 has earlier been reported by dilley et al @xcite .
all these results of transport properties measurements , we believe , are correlated to the anomalous fc response observed in and around the pe regime of ceru@xmath0 in various magnetic measurements @xcite .
however , ( to our knowledge ) no such report of history effects exists for v@xmath1si either in magnetic properties or in transport properties .
similar field - history effects are well known in the rfim systems@xcite .
most experimental information in this regard has been obtained from various diluted antiferromagnets in an applied magnetic field . in zfc mode ,
the diluted antiferromagnet is cooled ( in zero field ) through the zero - field neel transition temperature .
the resultant long range magnetic order is preserved when an external magnetic field is switched on at low temperatures .
this long range order , however , gradually decreases on heating the sample to the high temperature paramagnetic phase .
however , on cooling back now from the paramagnetic phase in the presence of the applied field ( i.e. in the fc mode ) , the sample develops a short range ordered domain state @xcite . the similarity with the vortex state in ceru@xmath0 and v@xmath1si is apparent here , namely the higher i@xmath5 in the fc vortex state of these systems clearly argues for a relatively more disordered fc vortex state . to draw the analogy further , we shall now deal with the metastability of the fc state .
it has been observed experimentally that the fc state in the rfims are unstable to field and temperature cycling below the phase boundary ; the fc state tend to get back the long range order through such cycling @xcite . to show the similar effects in the fc vortex state of ceru@xmath0 and v@xmath1si we subject the sample to field cycling after the initial field cooling experiment .
we have found that the intermediate field zero resistance state is readily destroyed by a subsequent field cycling through a small value ( of the order of few tens of mt ) and the corresponding zfc state flux - flow resistance is recovered ( see fig .
such metastability is not observed in the low h regime ( h @xmath6 0.8 t for ceru@xmath0 and [email protected] t for v@xmath1si ) and inside the pe regime ( 1.25t@[email protected] t for ceru@xmath0 and 3.48t@[email protected] t for v@xmath1si ) ; the zero resistance state is quite stable to any field cycling in these h - regime .
a continuous phase transition from an elastic vortex solid ( or bragg - glass ) to a plastic vortex solid ( or vortex - glass ) has been put forward as an explanation of the pe in various htsc materials @xcite . however , any zfc / fc history effect has not been associated with the pe in these materials so far . although it is widely accepted that the pe in ceru@xmath0 indicates a transition from a relatively ordered vortex solid to a disordered vortex solid@xcite the exact nature of this transition remains a matter of debate .
we have earlier suggested that the existence of a first order thermodynamic phase transition and the associated supercooling can explain various interesting features associated with pe in ceru@xmath0 including the history effects @xcite . here
the history effect is associated with the supercooling of the high field high temperature disordered vortex state ( with enhanced pinning ) across the transition line .
this picture gains further weight by our recent theoretical argument that the range of supercooling will be more while varying the temperature than while varying the field @xcite .
the experimental support in this regard already existed in magnetic studies of ceru@xmath0 , and further support is obtained through our present transport measurements .
there is a finite hysteresis in the field variation of the resistance in ceru@xmath0 between the ascending and the descending field cycle ( see inset of fig .
such a hysteresis in the isothermal field variation of r(h ) , however , is not very distinct for v@xmath1si .
as shown in fig.2 , the path dependence of r(h ) is very clear for both ceru@xmath0 and v@xmath1si when the vortex state prepared through the fc protocol . in spite of all these arguments
, we must point out here that a definite microscopic experimental evidence in support of ( or against ) a first order phase transition in ceru@xmath0 is yet to appear .
the analogy with the rfim systems continues here , since the question regarding the underlying phase transition is yet to be settled in the rfim systems also @xcite .
we have presented results of transport measurements in ceru@xmath0 which , in conjunction with our earlier magnetic measurements @xcite , show clear analogy between the vortex state of ceru@xmath0 around the pe regime and the rfim systems .
our study shows that like in rfim systems , the fc vortex state in ceru@xmath0 is relatively more disordered and metastable in character . in addition , we have shown here the existence of the same features in the well known superconductor v@xmath1si .
s. b. roy and p. chaddah , physica * c279 * ( 1997 ) 70 .
s. b. roy , p. chaddah and s. chaudhary , j. phys.:condensed matter , * 10 * ( 1998 ) 8327 .
s. b. roy , p. chaddah and s. chaudhary , solid st .
commun . * 109*,(1999 ) 427 .
r. modler et al , phys .
* 76 * ( 1996 ) 1292 .
f. steglich et al , physica * c263 * ( 1996 ) 498 .
s. b. roy and p. chaddah , j. phys.:condensed matter * 9 * ( 1997 ) l625 .
s. b. roy , p. chaddah and s. chaudhary , j. phys.:condensed matter , * 10 * ( 1998 ) 4885 .
a. d. huxley et al , physica * b223 - 224 * ( 1996 ) 169 .
n. r. dilley et al , phys . rev .
* b56 * ( 1997 ) 2379 .
j. a. mydosh , spin - glasses ( taylor and francis , 1993 ) . r. j. birgeneau , j. magn .
magn . mater . *
177 * ( 1998 ) 1
. m. isino et al , phys . rev .
* b38 * ( 1988 ) 4457 .
a. d. huxley et al , j. phys.:condens .
matter * 5 * ( 1993 ) 7709 .
r. corcoran et .
al . phys .
* 72 * ( 1994 ) 701 .
g. ravikumar et al , phys . rev .
* b57 * ( 1998 ) r11069 .
d. p. belanger , phase transitions * 11 * ( 1988 ) 53 .
d. p. belanger , a. r. king and v. jaccarino , solid st .
* 54 * ( 1985 ) 79
. b. khaykovich et al , phys .
* 76 * ( 1996 ) 2555 ; k. deligiannis et al , phys .
* 79 * ( 1997 ) 2121 ; d. giller et al , phys .
* 79 * ( 1997 ) 2542 .
p. chaddah and s. b. roy , phys
* b60*(1999 ) 11926 .
j. p. hill et al , phys .
* b55 * ( 1997 ) 356 ; q. feng et al , phys
* b55 * ( 1997 ) 370 . *
\(a ) resistance ( r ) vs field ( h ) plot for ceru@xmath0 obtained in the zfc mode at 5k with i@xmath2 = 100 ma .
open squares ( open triangles ) denote the data points in the ascending ( descending ) h cycle .
the inset shows the hysteresis in r(h ) between the ascending and the descending h cycle , at the onset of the pe regime .
( b ) resistance ( r ) vs field ( h ) plot for v@xmath1si at 14.5k with i@xmath2 = 85 ma . *
\(a ) metastable behaviour of r(h ) of ceru@xmath0 obtained in the fc mode at 5k with i@xmath2 = 100 ma .
solid circles denote r(h ) values obtained after field cooling in various h values .
filled triangles denote r(h ) values obtained after a field cycling of maximum 25 mt subsequent to the first fc measurement at the corresponding h. solid and dashed lines denote the r(h ) obtained in the isothermal zfc mode in the ascending and descending h respectively ( same as in fig .
( b ) metastable behaviour of r(h ) of v@xmath1si obtained in the fc mode at 14.5k with i@xmath2 = 85 ma .
solid circles denote r(h ) values obtained after field cooling in various h values .
filled triangles denote r(h ) values obtained after a field cycling of maximum 50 mt subsequent to the first fc measurement at the corresponding h. solid and dashed lines denote the r(h ) obtained in the isothermal zfc mode in the ascending and descending h respectively ( same as in fig . | * abstract * we present results of transport measurements showing distinct path dependence of the electrical resistance in the superconducting vortex state of single crystal samples of ceru@xmath0 and v@xmath1si . resistance measured in the vortex state of both the systems prepared by field cooling ( fc ) , indicates a relatively higher degree of disorder than when it is prepared by isothermal variation of field .
small oscillations of magnetic field modify the resistance in the fc state , highlighting the metastable nature of that state .
an analogy is drawn with the fc state of the random - field ising systems . |
we consider a group of agents arranged on the nodes of a graph who repeatedly play the prisoner s dilemma game against their immediate neighbors .
the players adapt their actions to the past behavior of their opponents by applying the so - called win - stay lose - shift strategy @xcite which , as the name suggests , consists in changing strategy whenever the payoff is deemed unsatisfactory .
this model has been studied in the artificial intelligence literature @xcite as a simple example of `` co - learning '' @xcite . on a finite connected graph
, it turns out that the system converges to the all - cooperate state the globally optimal state in finite time . in this respect , this instance of the iterated prisoner s dilemma ( ipd ) game on a graph provides an interesting example of a system learning to behave optimally by a mechanism that involves each agent applying independently a simple strategy or rule of thumb which takes into account only the latest actions of its immediate neighbors . for related work , see @xcite and references therein .
see also @xcite for the evolutionary perspective . in order to understand how persistent this `` emergence of cooperation ''
phenomenon is , it is crucial to analyze the rate of convergence to the all - cooperate state . where the convergence is rapid , one would expect to observe the optimal , cooperation state in a practical system based on similar dynamics . on the other hand , where the convergence is slow , one would rather expect that such a system would stagnate in a suboptimal , metastable state where a nonnegligible fraction of agents defect .
rates of convergence for ipd were studied in @xcite where the structure of the graph was shown to be a determining factor . in this paper , we show that ipd exhibits an exponentially slow convergence to cooperation on expander graphs and bounded - degree trees .
our result for bounded - degree trees is somewhat surprising . in particular ,
it should be compared to the behavior of global reversible dynamics on trees @xcite where the convergence is always rapid .
note however that this slow convergence on trees is not unprecedented .
notably , the contact process , a common model of infection , is slow to converge on trees when the infection rate is large .
see e.g. @xcite and references therein .
in fact , our proof suggests that ipd behaves very much like the contact process .
nevertheless , the analysis of non - reversible particle systems has been an open challenge in the last two decades and we hope that the results obtained here can shed some more light on how such systems can be tackled .
the proof of slow convergence we give here combines several ideas .
the main idea is to look at the process at the right space - time scaling .
this approach , commonly used in probability ( e.g. in the analysis of interacting particle systems @xcite ) , allows us to analyze the rough behavior of ipd
defection survives for long periods of time in zones that are densely populated by defectors .
the main technical difficulty is to control the dependencies between different regions and different times .
then the process is compared to a directed percolation process ( where the directed axis corresponds to the time axis in the original process ) . using contour arguments
we show that the directed percolation process survives for an exponential time .
see @xcite for background on directed percolation .
recall that the prisoner s dilemma game ( pd ) is a bimatrix game with the following payoff matrix for the row player ( and similarly for the column player ) : @xmath0 where @xmath1 and @xmath2 .
the first row ( column ) corresponds to the _ cooperate _ action and the second row ( column ) corresponds to the _ defect _ action .
the global or pareto optimum is for both agents to cooperate .
however , for any given action of the column player , it is always in the row player s advantage to defect ( and similarly for the column player ) . for an agent playing pd ,
a simple way to adapt to her opponent s behavior is the so - called win - stay lose - shift strategy ( wsls ) @xcite , also known as the pavlov strategy @xcite .
this works as follows .
every time the game is played , if the agent s payoff is one of the two smaller payoffs , i.e. @xmath3 or @xmath4 , then she switches her action in anticipation for the next round of play , otherwise she keeps the same action .
we now consider a repeated graphical version of pd which we will refer to as ipd .
let @xmath5 be a finite graph with @xmath6 .
each node , @xmath7 , is an agent to which we associate an action @xmath8 at time @xmath9 .
( as will become clear in later sections , it is easier to consider the continuous - time version of this problem . ) here @xmath10 stands for _ cooperation _ while @xmath11 stands for _ defection_. the initial state is @xmath12 for all @xmath13 .
the agents repeatedly play pd against their immediate neighbors in the graph through the following mechanism .
each edge @xmath14 has an exponential clock , i.e. we associate to each edge an independent poisson process @xmath15 where all inter - arrival times @xmath16 are independent exp(1 ) ( with the convention @xmath17 ) .
every time a clock rings , say at edge @xmath18 , the endpoint agents @xmath19 and @xmath7 play one round of pd using their respective actions @xmath20 and @xmath21 , assuming the clock rings at time @xmath22 .
then the two agents update their state using wsls . in other words , if a clock rings on edge @xmath18 at time @xmath22 , we witness the following transition for @xmath23 @xmath24 this defines a stochastic process for the state of the system @xmath25 with initial state the all - defect state , @xmath26 .
it is clear that , given the above allowed transitions , the system has a unique fixed point , the all - cooperate state @xmath27 .
in particular , if @xmath28 is a finite connected graph with @xmath29 , we have a.s . @xmath30 as @xmath31 .
the question of interest is : how long does it take to reach @xmath32 on a given graph .
it was shown by @xcite and previously conjectured in @xcite that the time to the emergence of cooperation depends crucially on the structure of the graph .
let @xmath33 be the stopping time at which @xmath34 reaches @xmath32 for the first time .
below , _ with high probability _
( w.h.p . ) means with probability @xmath35 where @xmath36 increases polynomially with @xmath37 . in @xcite ,
the following two results are proved .
let @xmath28 be a cycle on @xmath37 vertices . then w.h.p .
@xmath38 .
[ thm : complete ] let @xmath28 be the complete graph on @xmath37 vertices
. then w.h.p .
@xmath39 . given the previous theorems , it is natural to conjecture that the time to the emergence of cooperation is governed by the connectivity of the graph : a high connectivity , as in the complete graph , leads to slow convergence , while a low connectivity , as in the cycle , leads to fast convergence .
surprisingly , we refute this intuition with our main result .
[ thm : tree ] there is a constant @xmath40 so that for all @xmath37 there is a @xmath40-regular tree with @xmath37 vertices for which w.h.p .
@xmath41 for some @xmath42 depending only on @xmath40 . to prove this result , we study ipd on `` linear trees . ''
the main technical ingredient is a coupling with oriented percolation .
the proof of this theorem is given in section [ sec : tree ] .
although the connectivity conjecture turns out to be wrong in general , the following theorem , an extension of the complete graph result of @xcite , shows that the intuition is partly correct in one direction .
let @xmath28 be a graph with @xmath37 vertices .
let @xmath43 be two increasing functions of @xmath37 such that for all @xmath37 , @xmath44 .
define the @xmath45-expansion constant @xmath46 of @xmath28 as @xmath47 where @xmath48 is the set of edges between @xmath49 and @xmath50 , @xmath51 is the sum of the degrees of the nodes in @xmath49 , and @xmath52 is the cardinality of @xmath53 .
[ thm : expand ] let @xmath54 .
let @xmath43 be two increasing functions of @xmath37 such that for all @xmath37 , @xmath44 .
let @xmath28 be a graph with @xmath37 vertices such that @xmath55 .
then there is a constant @xmath56 ( depending only on @xmath57 ) such that w.h.p .
@xmath58 ( for @xmath37 large enough ) . in particular ,
if @xmath59 are linear in @xmath37 , the emergence of cooperation is exponentially slow .
this follows from a martingale argument similar to that used in @xcite which is detailed in section [ sec : expand ] .
note that in theorem [ thm : expand ] , in order to obtain slow convergence , it suffices to have large expansion for relatively small sets .
in particular , the theorem applies to expander graphs such as random regular graphs @xcite .
in this section , we analyze ipd on caterpillar trees of degree @xmath40 .
we define an _
( n , d)-caterpillar _ , denoted @xmath60 , to be a tree with the following property : the subtree induced by the internal nodes is a path containing @xmath37 nodes all of which have degree @xmath40 . see figure [ fig : cater ] .
our main result , theorem [ thm : tree ] , is that cooperation is slow to emerge on caterpillars .
the proof of theorem [ thm : tree ] follows from a series of stochastic domination arguments .
we now briefly outline the main steps of the proof . 1 . * star dynamics via biased random walk . *
the first step is to analyze the behavior of a single star .
the main point here is that it takes the star with @xmath40 leaves an exponential number of steps ( in @xmath40 ) to move from the all - defect state to the all - cooperate state .
this is proved by comparing the process to a biased random walk .
this comparison also shows that a star can go from a few defectors to linearly many in @xmath61 time with constant probability , and that a small linear fraction of defectors grows with high probability within @xmath61 steps .
moreover , these claims can be established even if one allows two of the nodes of the stars to have arbitrary values .
2 . * space - time scaling .
* we think of a star as defecting if at least @xmath62 of its leaves defect .
then , we consider triplets of adjacent stars and say that a triplet is defecting if at least one of its extremal stars is defecting .
( we actually work with triplets of stars rather than pairs to help control dependencies . )
we scale time by looking at the process every @xmath61 steps .
the random walk argument of the previous point allows to show that defecting stars have a high probability at least @xmath63of remaining defectors after the @xmath61 time window .
moreover , a defecting star has a @xmath64 probability of `` infecting '' neighboring stars during that time . by iterating these observations @xmath61 times yet another time scaling we show that a defecting triplet has a probability @xmath65 of `` infecting '' a neighboring triplet .
( neighboring triplets are actually intersecting . )
* percolation .
* we may now look at the space - time diagram of defecting triplets and show that it dominates a directed percolation with probability @xmath66 for edges to be open .
the time axis of the original process corresponds to the direction of propagation in the percolation process .
finally , a contour argument allows to conclude that this percolation survives for a time which is exponential in @xmath37 , thus proving that the convergence time of ipd on the caterpillar is itself exponential in @xmath37 .
let @xmath67 .
this graph is made of @xmath37 copies of @xmath68 ( i.e. stars of degree @xmath40 ) .
let @xmath69 be any star in @xmath28 .
denote the root @xmath70 and the leaves @xmath71 .
a crucial property of stars is that cooperation is slow to emerge on them .
this follows from our next result .
we single out nodes @xmath72 and @xmath73 , which are defined to be the two nodes that @xmath69 shares with its neighboring stars .
( in the case of extremal stars , we just pick an arbitrary node in addition to the node shared with the next star . )
we call @xmath72 and @xmath73 the external vertices .
we use the following notation : @xmath74 .
[ lem : star ] consider the ipd chain @xmath75 on @xmath76 with @xmath77 .
let @xmath69 be an arbitrary star in @xmath28 with nodes denoted @xmath78 ( @xmath70 being the root , and @xmath72 and @xmath73 being the external vertices ) .
let @xmath79 be a positive integer and @xmath80 be three increasing functions of @xmath40 with @xmath81 and @xmath82 satisfying @xmath83 for all @xmath40 .
let the initial configuration be as follows .
on @xmath69 , nodes @xmath84 through @xmath85 are @xmath10 and nodes @xmath86 through @xmath40 are @xmath11 . on all other nodes , including the root and external vertices of @xmath69 , the initial state is arbitrary .
define @xmath87 let @xmath88 be the first time @xmath89 .
let @xmath90 , @xmath91 , @xmath92 , and @xmath93 .
then , we have @xmath94 \leq 2^{-\delta_1 } + \rho^{-\sqrt{\mu}/2}(\sqrt{2})^{\delta_2 } + 2^{-\mu/2}.\ ] ] moreover , this bound applies simultaneously on all stars independently from each other ( possibly with different choices of @xmath95 s ) .
* proof : * for this argument , we restrict ourselves to what happens on @xmath69 and do not refer to any event involving the rest of @xmath28 .
we call a leaf edge with leaf state @xmath11 a @xmath11-edge , and similarly for @xmath10 .
the behavior of @xmath96 depends on the state at the root of @xmath69 .
when @xmath97 , nothing happens until a @xmath11-edge is picked at which time @xmath98 becomes @xmath11 itself . on the other hand ,
when @xmath99 , either a @xmath10-edge is chosen in which case @xmath96 may go up by 1 ( or stay the same if @xmath72 or @xmath73 is picked ) , or a @xmath11-edge is chosen in which case @xmath96 may go down by 1 ( or stay the same if @xmath72 or @xmath73 is picked ) and @xmath98 becomes @xmath10 .
ignore the updates where nothing changes , i.e. when an edge @xmath100 is chosen . in any configuration
satisfying @xmath101 , there are at least @xmath102 edges whose updates change the configuration .
let @xmath103 the number of such updates in time @xmath79 .
then it follows that @xmath103 is larger than a poisson with mean @xmath93 . from the moment generating function of the poisson distribution ( see e.g. @xcite ) , we have the following @xmath104 = { \mathbb{p}}[e^{\mu - q } \geq e^{\mu - \sqrt{\mu } } ] \leq \frac{{\mathbb{e}}[e^{\mu - q}]}{e^{\mu - \sqrt{\mu } } } \leq \frac{e^{\mu } e^{\mu ( e^{-1 } - 1)}}{e^{\mu - \sqrt{\mu } } } \leq 2^{-\mu/2}.\ ] ] assume the event @xmath105 holds .
also , note that at most one out of 2 steps have @xmath97 .
( remember that we ignore @xmath100 updates . )
ignore the times with @xmath97 as well , what remains is an asymmetric random walk ( or rather a birth - and - death chain ) which does at least @xmath106 steps before time @xmath79 . to bound the probability that @xmath96 goes up or down , we use the fact that the chain starts with @xmath107 @xmath11 s and is stopped when it reaches either @xmath102 or @xmath108 @xmath11 s . by assumption
, the probability that @xmath96 goes up when @xmath99 is at least @xmath109 .
consider the walk @xmath110 on @xmath111 started at @xmath112 which goes up with probability @xmath113 and goes down with probability @xmath114 .
let @xmath115 be the time at which @xmath116 reaches @xmath95 . for convenience , we assume that the process @xmath117 is defined on all of @xmath118 ( even though outside the interval @xmath119 $ ] the bounds used are not valid ) . then , @xmath120
\leq { \mathbb{p}}[t'_{g_2 } \geq ( t'_{g_0}\land \sqrt{\mu}/2 ) ] \leq { \mathbb{p}}[t'_{g_2 } \geq t'_{g_0 } ] + { \mathbb{p}}[t'_{g_2 } \geq \sqrt{\mu}/2].\ ] ] by standard martingale results ( see e.g. @xcite ) , we have
@xmath121 = \frac{\phi(\delta_2 ) - \phi(0)}{\phi(\delta_2 ) - \phi(-\delta_1)},\ ] ] where @xmath122 so , @xmath121 = \frac{1 - 2^{-\delta_2}}{2^{\delta_1 } - 2^{-\delta_2 } } \leq 2^{-\delta_1}.\ ] ] we also have @xmath123 = \left(\frac{1 - \sqrt{1 - 4p(1 - p)\rho^2}}{2 ( 1-p)\rho}\right)^{\delta_2}.\ ] ] the choice @xmath92 gives @xmath123 = \left(\sqrt{2}\right)^{\delta_2}.\ ] ] by markov s inequality , @xmath124 = { \mathbb{p}}[\rho^{t'_{g_2 } } \geq \rho^{\sqrt{\mu}/2 } ] \leq \rho^{-\sqrt{\mu}/2}\left(\sqrt{2}\right)^{\delta_2}.\end{aligned}\ ] ] finally , putting everything together , we get @xmath125
. the independence of the bound at each star in @xmath28 comes from the fact that we use only events involving leaf edges of @xmath69 .
@xmath126 the following corollary corresponds to the case where a star has initially only a few @xmath11 s .
the result below implies that after @xmath127 steps the star has @xmath128
@xmath11 s with positive probability .
[ cor : birth ] in the setup of lemma [ lem : star ] , let @xmath129 , @xmath130 and @xmath81 .
then , for @xmath131 and @xmath40 ( constant ) large enough , we have @xmath132 \leq \frac{2}{3}.\ ] ] the following corollary implies that a star with @xmath128 @xmath11 s still has @xmath128 @xmath11 s after @xmath61 steps , with high probability .
[ cor : survival ] in the setup of lemma [ lem : star ] , let @xmath133 , @xmath81 , @xmath134 , and @xmath135 .
then , @xmath136 \leq 2^{-d/12}.\ ] ] the following corollary implies that a star with @xmath62 @xmath11 s reaches @xmath137 @xmath11 s after @xmath61 steps , with high probability .
[ cor : boosting ] let @xmath138 be a positive integer , not depending on @xmath40
. in the setup of lemma [ lem : star ] , let @xmath81 , @xmath139 , and @xmath140 .
then , for @xmath131 and @xmath40 large enough , we have @xmath132 \leq 3\ , 2^{-d/20 } \leq 2^{-d/21}.\ ] ] the next step in the proof of theorem [ thm : tree ] is to make the connection between ipd and oriented percolation . here
we show how a triplet of stars dominates the building block of a percolation lattice .
we use the following oriented percolation . consider four adjacent vertices of the regular lattice @xmath141 , say @xmath142 , @xmath143 , @xmath144 and @xmath145 .
assume the nodes are connected by four directed edges : @xmath146 , @xmath147 , @xmath148 , and @xmath149 .
see figure [ fig : perc1 ] .
each edge is _ open _ with respective probability @xmath150 , @xmath151 , @xmath152 , and @xmath153 .
the vertices have a state , denoted respectively @xmath154 , @xmath155 , @xmath156 , @xmath157 , which takes its value in @xmath158 .
the state @xmath72 `` travels''along the open edges , i.e. if @xmath159 is an open edge and the state at @xmath19 is @xmath72 then the state at @xmath7 is also @xmath72 .
a vertex is in state @xmath72 if and only if it is the terminal vertex of an open edge with initial vertex in state @xmath72 .
we denote this four - node graph @xmath160 .
now consider any triplet of adjacent stars inside @xmath161 .
denote the stars @xmath162 , @xmath163 , with corresponding edges @xmath164 and vertices @xmath165 , with the label @xmath70 corresponding to the root .
we have the correspondence @xmath166 and @xmath167 .
we denote this subgraph which is a copy of @xmath168@xmath69 .
we are interested in the number of @xmath11 s on each star , excluding nodes @xmath70 , @xmath72 , and @xmath73 of each star , which we denote @xmath169 .
the detailed behavior of @xmath170 is rather intricate .
we simplify the process by projecting it to a smaller space .
let @xmath171 = \left\ { \begin{array}{ll } 1 , & \mathrm{if\ } n > d/4 - 2,\\ 0 , & \mathrm{if\ otherwise . }
\end{array } \right.\ ] ] consider the random vector @xmath172 , \sigma_d[n_m^{(1 ) } ] , \sigma_d[n_0^{(3 ) } ] , \sigma_d[n_m^{(3)}]\right),\ ] ] for some @xmath173 .
the following lemma shows that for an appropriate choice of @xmath174 , @xmath175 , @xmath176 , @xmath152 , and @xmath153 , the vector @xmath177 stochastically dominates @xmath178 defined by the percolation above ( with @xmath179 and @xmath180 ) .
[ lem : triplet ] consider the ipd chain @xmath75 on @xmath76 with @xmath77 .
let @xmath69 be an arbitrary triplet of adjacent stars in @xmath28 .
let @xmath181 , @xmath182 , and @xmath183 .
then , for any initial configuration and @xmath184 such that @xmath179 and @xmath180 , we have that @xmath185 stochastically dominates @xmath186 for @xmath40 ( constant ) large enough
. moreover , the domination holds for any number of ( edge-)nonintersecting triplets simultaneously independently from each other .
* proof : * the argument ignores any event outside @xmath69 .
we consider three cases .
\1 ) * case @xmath187 . * in that case , we have @xmath188 , which is of course dominated by @xmath189 .
\2 ) * case @xmath190 .
* we use corollaries [ cor : survival ] and [ cor : boosting ] , which we apply to stars @xmath72 and @xmath84 independently . consider star @xmath72 .
we first go through a `` boosting '' phase where we let @xmath191 drift from @xmath192 to @xmath193
. then we compute the probability that @xmath191 stays above @xmath192 for the remaining time .
* phase 1 . * for the boosting phase , we apply corollary [ cor : boosting ] .
the probability of remaining below @xmath194 is at most @xmath195 .
* phase 2 . *
the time remaining after boosting is of course at most @xmath174 . in time
@xmath174 , there is a poisson number of steps , say @xmath196 , with mean @xmath197 ( including the steps where nothing happens ) . from the moment generating function of the poisson distribution ( see e.g. @xcite ) , we have the following @xmath198 = { \mathbb{p}}[e^{q ' } \geq e^{d^2m^2 } ] \leq \frac{{\mathbb{e}}[e^{q'}]}{e^{d^2m^2 } } \leq \frac{e^{dm(e - 1)}}{e^{d^2m^2 } } \leq 2^{-d^2m^2/2}.\ ] ] assuming @xmath194 was reached and that there remains at most @xmath199 discrete steps , we get that there are at most @xmath200 crossings of the interval @xmath201 $ ] by the process @xmath191 . by corollary [ cor : survival ] , every time @xmath202 , there is a probability of at least @xmath203 of coming back to @xmath194 before hitting @xmath204 .
the probability that any of @xmath199 attempts at crossing @xmath201 $ ] succeeds is at most at most @xmath205 which implies @xmath206 \leq d^2 m^2 2^{-d/12 } + 2^{-d^2m^2/2 } + 2^{-d/21 } \leq 2^{-d/22},\ ] ] for @xmath40 large enough .
stochastic domination of the oriented percolation follows directly .
\3 ) * case @xmath207 , @xmath208 .
* ( the symmetric case is analyzed similarly . )
we divide the time window in two phases . for the first phase , we compute the probability that defection `` spreads '' from star @xmath72 to star @xmath84 . for the second phase ,
we compute the probability that stars @xmath72 and @xmath84 remain in or reach state @xmath72 respectively
. * phase 1 . *
it is easy to see that , in any initial configuration satisfying @xmath207 , @xmath208 , six steps ( or less ) suffice to reach a configuration with @xmath209 .
the probability that the first six steps taken by ipd satisfy this property call that event @xmath210is at least @xmath211 .
let @xmath212 be the number of steps until time @xmath213 .
then , @xmath214 \leq 2^{-m/4},\ ] ] by a calculation similar to that in lemma [ lem : star ] . * phase 2 .
* we condition on @xmath215 .
consider first star @xmath72 .
whether or not @xmath210 is realized , at the beginning of phase 2 , we have @xmath216 .
we are back in the situation of case 2 ) , except that the time left is only at least @xmath213 . by the same calculation
, we obtain that the probability that @xmath217 is @xmath70 is at most @xmath218 for @xmath40 large enough .
consider now star @xmath84 .
let @xmath219 be the number of discrete steps left on star @xmath84 .
the time remaining is at least @xmath213 .
it follows from corollary [ cor : birth ] that @xmath220 reaches @xmath194 before the end of the time window with probability at least @xmath221 for @xmath40 large enough .
once @xmath194 is reached , we are back to phase 2 of case 2 ) .
it follows that on @xmath215 the probability that @xmath222 is at least @xmath223 .
note that on @xmath215 , the bounds on star @xmath72 and @xmath84 are independent .
it is then easy to check that stochastic domination of the oriented percolation holds .
@xmath126 we further simplify the chain by stacking up the construction in the previous lemma and projecting once more to a smaller space .
for this , we consider a different percolation model on @xmath141 .
see figure [ fig : perc2 ] .
let @xmath224 be the directed graph made of three nodes @xmath225 with two edges @xmath226 , @xmath227 .
the edges are open with probability @xmath228 , @xmath229 respectively .
the nodes have state @xmath230 respectively with value in @xmath231 .
the percolation works as before with state @xmath72 `` traveling '' along open edges .
consider again ipd on an arbitrary triplet of stars @xmath69 of @xmath28 .
redefine the vector @xmath177 by taking instead @xmath172 , \sigma_d[n_{im}^{(1 ) } ] , \sigma_d[n_0^{(3 ) } ] , \sigma_d[n_{im}^{(3)}]\right),\ ] ] for some @xmath232 and @xmath174 as in lemma [ lem : triplet ] .
we use the following notation : @xmath233 .
[ lem : tower ] consider the ipd chain @xmath75 on @xmath76 with @xmath77 .
let @xmath69 be an arbitrary triplet of adjacent stars in @xmath28 .
let @xmath181 , @xmath234 , and @xmath235 .
then , for any initial configuration and @xmath236 such that @xmath237 , we have that @xmath185 stochastically dominates @xmath238 for @xmath40 ( constant ) large enough .
moreover , the domination holds for any number of ( edge-)nonintersecting triplets simultaneously independently from each other .
* proof : * the argument ignores any event outside @xmath69 .
the proof works by stacking up @xmath239 copies of @xmath160 and applying lemma [ lem : triplet ] .
consider again @xmath141 .
we define a _
i - tower _ , denoted @xmath240 , to be the graph on nodes @xmath241 where each set of four nodes of the form @xmath242 induces a copy of @xmath160 with the same values of @xmath243 as in lemma [ lem : triplet ] .
the node states are denoted @xmath244 . by applying repeatedly
lemma [ lem : triplet ] , we get that , if @xmath245 , then @xmath185 stochastically dominates @xmath246 , so it suffices to show that the latter dominates @xmath247 .
the case @xmath248 is trivial .
so assume @xmath249 .
then , the subcase @xmath250 dominates the subcase @xmath251 so it suffices to consider the latter . without loss of generality ,
let @xmath207 and @xmath208 .
the probability that at least one upwards edge in @xmath240 is closed is at most @xmath252 for @xmath40 large enough .
the probability that no up - right edge is open is at most @xmath253 for @xmath40 large enough .
therefore , @xmath254 \geq 1 - 2^{-d/32},\ ] ] for @xmath40 large enough .
but note that @xmath255 = ( 2^{-d/100})^2 = 2^{-d/50 } \geq 2^{-d/32}.\end{aligned}\ ] ] so we have domination .
@xmath126 we conclude the proof of theorem [ thm : tree ] by showing that the ipd chain at intervals of time @xmath256 dominates a standard percolation model and that in turn the latter model percolates at an exponential distance from its bottom nodes
. for convenience , assume @xmath37 is of the form @xmath257 for some positive integer @xmath258 .
( the reason for this choice will be clear below .
see also figure [ fig : period ] . )
consider the following sublattice of @xmath141 , @xmath259 where @xmath260 is a positive integer that will be fixed below .
consider the directed graph @xmath261 with node set @xmath262 and edge set @xmath263 see figure [ fig : perc3 ] for an illustration .
each edge has probability @xmath264 of being open where @xmath264 is set below .
we consider the percolation process on @xmath265 and denote the states @xmath266 .
let @xmath75 be the ipd chain on @xmath267 and denote @xmath268 the number of @xmath11 s on star @xmath269 at time @xmath22 , excluding the external nodes .
we consider the following projection of @xmath75 .
let @xmath270 and let @xmath271 where @xmath272\lor\sigma_d\left[n_{jim}^{(\mu(i , j ) + 1)}\right],\ ] ] where @xmath239 and @xmath174 are as in lemma [ lem : tower ] . see figure [ fig : period ] .
we show first that @xmath273 dominates @xmath274 .
[ lem : perc ] consider the ipd chain @xmath75 on @xmath76 with @xmath77 .
let @xmath181 , @xmath234 , and @xmath275 .
let @xmath276 ( the all-@xmath11 state ) and let @xmath277 for all even @xmath269 s .
then , we have that @xmath273 stochastically dominates @xmath274 for @xmath40 ( constant ) large enough .
* proof : * this actually follows immediately from lemma [ lem : tower ] .
@xmath126 finally , the next lemma concludes the proof of theorem [ thm : tree ] .
let @xmath274 be defined as above with @xmath275 and let @xmath277 for all even @xmath269 s .
let @xmath278 .
assume that @xmath279 and that @xmath260 is even
. then @xmath280 \leq 2^{-(d/1000)n},\ ] ] for @xmath40 ( constant ) large enough . *
proof : * we use a standard duality argument . for more details , see @xcite .
first we modify the percolation lattice @xmath265 , which we now call the primal lattice and still denote @xmath265 . to each edge , we add another edge , reversed , with associated probability of being open @xmath70 .
we now define the dual lattice .
let @xmath281 consider the directed graph @xmath282 with node set @xmath283 and edge set @xmath284 superimpose @xmath265 on top of @xmath285 and notice that to each edge of @xmath285 corresponds an edge of @xmath265 which is rotated @xmath286 clockwise .
see figure [ fig : dual ] .
we couple the two lattices so that an edge in @xmath285 is closed if and only if the corresponding edge in @xmath265 is open .
it is not hard to see that there is an open path from level @xmath70 to level @xmath260 in @xmath265 if and only if there is no open path from the right boundary to the left boundary in @xmath285 .
so it remains to compute an upper bound on the latter probability .
fix any two boundary nodes in @xmath287 , say @xmath288 and @xmath289 for some @xmath290 .
the number of paths of length @xmath291 between @xmath292 and @xmath293 is at most @xmath294 .
each such path makes @xmath295 more moves to the left than to the right .
in particular , the number of moves to the left is at least @xmath296 .
moreover , each edge going to the left has a probability @xmath297 of being open .
so the probability that there is a path between @xmath292 and @xmath293 ( which we denote @xmath298 ) is at most @xmath299 \leq \sum_{l = n'-1}^{+\infty } 3^l ( 1 - p')^{l/2 } \leq \frac{(3 \ , 2^{-d/200})^{\frac{n-1}{2}}}{1 - 3\ , 2^{-d/200}},\end{aligned}\ ] ] for @xmath40 large enough .
there are at most @xmath300 pairs of boundary nodes so by the union bound @xmath301 \leq t^2 \frac{(3 \ , 2^{-d/200})^{\frac{n-1}{2}}}{1 - 3\ , 2^{-d/200 } } \leq 2^{-(d/1000)n},\end{aligned}\ ] ] for @xmath40 large enough .
for this section , we consider the discrete - time version of the chain .
that is , at every time step , we pick one edge uniformly at random and update the actions at the endpoints of that edge .
equivalently , we look at the discrete - time chain embedded in @xmath75 by stopping the chain every time a clock rings .
also , since we are looking for a lower bound on @xmath33 , we can speed up the chain by picking only those edges with at least one @xmath11 endpoint . denote the discrete - time sped - up chain @xmath302 .
the proof of theorem [ thm : expand ] is based on the following geometric observation .
let @xmath303 be the set of nodes defecting at time @xmath304 and denote @xmath305 .
at the next update , @xmath306 goes down by 2 if we pick an edge `` inside '' @xmath303 and it goes up by 1 if we pick an edge on the `` boundary '' of @xmath303 . therefore , if the boundary of @xmath303 is more than twice as big as the inside of @xmath303 , on average the chain moves away from the fixed point @xmath32 .
* proof of theorem [ thm : expand ] : * let @xmath307 with @xmath308 . note first that @xmath55 implies @xmath309 let @xmath310 such that @xmath311 .
then @xmath312 which implies @xmath313 therefore there is an @xmath314 such that if @xmath315 , then @xmath316 let @xmath317^{1/3 } > 1.\ ] ] it is easy to check that @xmath318 therefore , @xmath319 is a bounded nonnegative supermartingale on @xmath320 . using the optional sampling theorem as in @xcite , it follows that the probability of @xmath306 crossing the interval @xmath321 $ ] is less than @xmath322 for @xmath37 large enough
. the theorem immediately follows .
the work presented here leads naturally to the following questions : 1 .
is there a @xmath40 ( constant ) such that for all @xmath37 large enough and for all trees of minimum degree @xmath40 with @xmath37 nodes , the emergence of cooperation is exponentially slow ? 2 . what is a good criterion for fast emergence of cooperation in this setup ?
is the line and its appropriately defined
variants the only graphs on which the convergence to all - cooperation is fast ?
the first author acknowledges the support of a miller fellowship in statistics and computer science , u.c .
berkeley , a sloan fellowship in mathematics and nsf grants dms-0504245 and dms-0528488 .
the second author is supported by cipres ( nsf itr grant # nsf ef 03 - 31494 ) , fqrnt , nserc and a love fellowship .
the second author also thanks martin nowak and the program for evolutionary dynamics at harvard where part of this work was done .
n. berger , c. kenyon , e. mossel , and y. peres .
glauber dynamics on trees and hyperbolic graphs . , 131(3):311340 , 2005 .
extended abstract by kenyon , mossel and peres appeared in proceedings of 42nd ieee symposium on foundations of computer science ( focs ) 2001 , 568578 .
kittock , j.e .
emergent conventions and the structure of multi - agent systems . in : _ 1993 lectures in complex systems : proceedings of the 1993 complex systems summer school _ , nadel , l. and stein , d , eds . vol .
vi of _ santa fe institute studies in the sciences of complexity lecture _
, santa fe institute , addison - wesley . | we consider a group of agents on a graph who repeatedly play the prisoner s dilemma game against their neighbors .
the players adapt their actions to the past behavior of their opponents by applying the win - stay lose - shift strategy . on a finite connected graph , it is easy to see that the system learns to cooperate by converging to the all - cooperate state in a finite time .
we analyze the rate of convergence in terms of the size and structure of the graph .
[ dyer et al . , 2002 ] showed that the system converges rapidly on the cycle , but that it takes a time exponential in the size of the graph to converge to cooperation on the complete graph .
we show that the emergence of cooperation is exponentially slow in some expander graphs .
more surprisingly , we show that it is also exponentially slow in bounded - degree trees , where many other dynamics are known to converge rapidly .
* keywords : * games on graphs , learning , prisoner s dilemma game , win - stay lose - shift , oriented percolation , emergence of cooperation . |
the flavour structure of nucleons is described in terms of parton distribution functions ( pdf ) .
the unpolarized pdf @xmath0 gives essentially the probability to find a parton with momentum fraction @xmath1 in a nucleon .
the polarized pdf s describe the spin structure of the nucleon .
in particular the longitudinal pdf @xmath2 gives the helicity distribution of partons in the nucleon .
unpolarized pdf s have been measured in great detail over a wide range in @xmath1 and @xmath3 .
the information on polarized pdf s is much more sketchy .
a reasonably accurate picture of the quark helicity structure has emerged in recent years , but other aspects of the spin structure remain virtually untouched . among these
the lack of experimental information on the gluon polarization and the transverse spin structure of the nucleon are particularly obvious .
several experiments are gearing up to address these issues in the near future .
this paper reports on some recent results obtained by the hermes experiment in connection with the longitudinal spin structure of the nucleon , which was the main topic for the hermes run - i ( 1995 - 2000 ) .
the plans for the run - ii ( 2002 - 2006 + ) are discussed in the second half of this contribution . here
the main interest lies in exclusive reactions using a new recoil detector , and the transverse spin structure of the nucleon .
hermes is a fixed target experiment @xcite on the 27.6 gev hera-@xmath4 ring .
the electrons in the ring are transversely polarized by the sokolov - ternov effect , and longitudinal polarization is achieved by a spin rotator in front of the hermes target .
a second rotator behind the experiment restores the transverse polarization .
typical beam polarizations are between 50 and 60% .
the polarization of the beam is continuously monitored by two independent compton backscattering polarimeters .
the beam traverses the hermes target consisting of a 40 cm long windowless storage cell which confines the injected polarized target gas to the region around the beam .
the polarized atomic h and d are provided by an atomic beam source using stern - gerlach separation . a small sample of the target gas is continuously analyzed in a breit - rabi polarimeter .
typical polarizations are 85% . in 1996 and 1997
data were taken on a polarized proton target , while in 1998 , 1999 and 2000 a large data sample was collected on a polarized deuterium target .
the scattered electrons , and some of the hadrons produced in the reaction , are detected in the hermes spectrometer .
this is a typical forward magnetic spectrometer consisting of two symmetric halves above and below the plane of the accelerator .
scattered leptons and produced hadrons are detected and identified within the angular acceptance of @xmath5 mrad horizontally and @xmath6 mrad vertically .
the hadron - lepton separation is done on the basis of the signals in a transition radiation detector , an electromagnetic calorimeter and a pair of scintillator hodoscopes where the second one is preceded by a lead sheet preshower .
the identification of pions , kaons and ( anti)protons is performed since 1998 with a ring imaging cherenkov detector ( rich ) using clear aerogel and c@xmath7f@xmath8 gas as radiators @xcite .
the hadrons can be separated over almost the full momentum range of hermes , i.e. between 2 and 15 gev . before 1998
a threshold erenkov was used to give pion identification in a limited momentum range .
the typical kinematic domain for the analysis of polarized dis in hermes is @xmath9 gev@xmath10 , @xmath11 , with @xmath3 the ( negative ) momentum transfer squared , and @xmath12 the bjorken scaling variable .
to avoid regions with too large contributions of resonance production or radiative corrections further conditions @xmath13 gev , @xmath14 are usually imposed , with @xmath15 the invariant mass ( squared ) of the photon - nucleon system and @xmath16 the fraction of the initial beam energy @xmath17 transferred . in semi - inclusive reactions where in addition to the scattered lepton
also a produced hadron is detected the fraction of the energy of the virtual photon carried by the hadron is given by @xmath18 . to avoid the region with predominantly target fragmentation a cut at @xmath19 is usually used .
most of the information on the pdf s has up to now come from inclusive dis : experiments where only the scattered lepton is detected . fig .
[ fig1 ] shows the world data on the polarized structure function @xmath20 for deuterium . to it is obvious from this figure that all experiments are basically in agreement with each other
the recent preliminary hermes data considerably improve the statistical accuracy .
however , there is a difference of about one order of magnitude in the scale of the various experiments , with the smc results being taken at much higher beam energy than the data from slac and hermes .
the fact that the ratio is similar in the different experiments thus implies an evolution for @xmath21 that is very similar to the one for @xmath22 . just as was done succesfully in the case of unpolarized structure functions one can analyse the scaling violations observed in @xmath23 for proton and neutron targets , and deduce information on the polarization of the different quark flavours and the gluon polarization
such a nlo qcd analysis was recently undertaken by hermes @xcite .
the results confirm the facts already known about the quark polarization : the quark spin contributes only 20 - 30% to the spin of the nucleon ; most of this contribution comes from the valence quarks ; the light quark sea is only little ( and negatively ) polarized .
the gluon polarization that can be derived from the analysis of the presently available data is shown in fig .
[ fig2 ] .
to this figure illustrates that the present data do not really constrain the gluon polarization .
the only strong suggestion is that it should be positive , but no value can be given .
more direct methods to determine the gluon contribution to the nucleon spin are necessary .
this is one of the aims of the compass experiment at cern which is now starting up .
also the spin physics programme at rhic intends to determine the gluon polarization independently .
the hermes experiment was specifically designed to perform accurate measurements of semi - inclusive reactions , where apart from the scattered lepton also some of the produced hadrons are detected .
the idea is that these hadrons contain extra information on the quark that took part in the scattering process .
this technique of flavour tagging allows the determination of the polarization of individual quark flavours directly from the spin asymmetries observed in hadron production .
the quantity of interest here is the photon - nucleon asymmetry @xmath24 when a hadron of type @xmath25 is produced , which can be derived from the experimental semi - inclusive spin asymmetry @xmath26 : @xmath27 where @xmath28 is the virtual photon depolarization , @xmath29 and @xmath30 are kinematic factors , and @xmath31 are the number of semi - inclusive events ( properly taking into account polarization of target and beam , and relative luminosity ) for target polarization parallel ( antiparallel ) to the beam polarization .
an example of the asymmetries obtained at hermes is given in fig .
similar asymmetries were derived for positive and negative pions and on a proton target .
to a striking feature of the asymmetries in this figure is the fact that the asymmetry for the @xmath32 is compatible with zero .
since this is an all - sea object this already indicates that the polarization of the quark sea will be very small . in leading order qcd and assuming factorization of the cross section one can write this asymmetry in terms of the pdf and fragmentation functions : @xmath33 in the last equation the _ purities _
@xmath34 are introduced .
these are spin - independent quantities which give the probability that a hadron of type @xmath25 observed in the final state came from a struck quark of flavour @xmath35 . in that sense
they are the inverse of fragmentation functions .
the purities depend on the unpolarized quark densities and the fragmentation functions .
the former were measured to high precision in unpolarized dis experiments .
information on the fragmentation functions is , however , less precise .
there is some information for pions , but those data are taken at quite different kinematics than relevant for hermes .
hence , the purities were calculated using a monte carlo simulation of the entire scattering process .
standard unpolarized parton distribution parametrizations @xcite were used , while the fragmentation was modelled in the lund string model as implemented in jetset @xcite .
the lund model fully describes the fragmentation from the current as well as the target fragmentation region .
the parameters of the model were tuned to fit the hadron multiplicities measured at hermes in order to achieve a good description of the fragmentation process at our kinematics .
the resulting purities also include effects of the acceptance of the spectrometer .
the analysis applies eq .
[ a1h ] in matrix form : @xmath36 where the elements of @xmath37 are the measured inclusive and semi - inclusive ( born ) asymmetries , the vector @xmath38 contains the quark ( and anti - quark ) polarizations and the matrix @xmath39 contains as elements the effective integrated purities for the proton and neutron targets .
this is an over - constrained system of equations which is solved by a minimisation procedure .
the purity formalism has been used in the hermes analysis @xcite to make a flavour decomposition of the quark polarizations for @xmath40 and @xmath41 . a symmetric strange sea polarization @xmath42 was assumed .
the measured asymmetries were integrated over the @xmath43-range from 0.2 to 0.8 , and over @xmath44gev@xmath10 .
the resulting spin densities are shown in fig .
[ 5qdens ] . to
the polarization of the @xmath45-quark is positive over the entire measured range in @xmath1 , with the largest polarization at high @xmath1 where the valence quarks dominate .
the polarization of the down - quark is negative and also most important in the region of the valence quarks .
the polarization of the light sea flavours @xmath46 and @xmath47 , and the polarization of the strange sea are consistent with zero .
the data are compared to polarized pdf s derived from fits to inclusive data .
good agreement is found between the spin densities directly measured here and the results of nlo qcd analyses , in particular for the valence distributions .
the tendency of the parameterizations to yield a negative sea polarization is not confirmed by the data , not is it ruled out with the present statistical accuracy . the flavour symmetry of the unpolarized light sea is known to be broken .
this is experimentally well established through a violation of the gottfried sum rule .
several models which give a good description of this symmetry breaking in the unpolarized sector also predict a sizeable symmetry breaking in the polarized sector .
in particular , most models predict that @xmath48 .
the flavour asymmetry derived from the present measurement is shown in fig .
[ du - db ] .
to the present data do not favour a very strong flavour symmetry breaking .
however , the statistical uncertainties are still rather large and further measurements of this quantity will be of great interest .
when analyzing the experimental inclusive spin asymmetries for deuterium in order to deduce the structure function @xmath49 one has to take into account a small effect due to possible tensor polarization in this spin 1 target .
this is connected with the presence in a spin 1 target of an additional tensor polarized structure function @xmath50 which relates to the other structure functions @xmath22 and @xmath21 as : @xmath51 \label{b1}\end{aligned}\ ] ] where @xmath52 and @xmath53 denote the probability of finding a quark with helicity parallel to that of the nucleon and @xmath54 is the probability to find a quark with momentum fraction @xmath1 in a nucleon with the target in a helicity state 0 .
@xmath50 thus measures the difference in parton distributions between a @xmath55 and a @xmath56 target .
the hermes target is capable of producing highly tensor polarized deuterium .
this was done during a short run in 2000 with an average polarization of 83% .
the tensor polarized structure function can be found from the experimental tensor asymmetry : @xmath57 the preliminary hermes result for this asymmetry is shown in fig .
[ at_fig ] , while the deduced tensor structure function @xmath58 ( trivially related to @xmath50 through : @xmath59 ) is shown in fig .
[ b2 ] . to to
the small values found for the tensor asymmetry ( of the order 1% ) show that the effect on the determination of the helicity structure function @xmath21 can safely be neglected .
the actual tensor structure function @xmath60 is significantly different from zero , in particular at lower valus of @xmath1 .
it is in broad agreement with the model calculations of ref
. @xcite .
exclusive dis reactions , where the target nucleon remains in or close to its ground state , can be described in terms of the generalized parton distributions ( gpd ) that were introduced a few years ago .
these gpd s form a natural off - forward extension of the standard parton distribution functions which are well determined from ( semi-)inclusive dis reactions .
they form a connection between the pdf s and the form - factors of hadrons .
the cleanest example of the appearance of gpd s in an exclusive reaction is that of deeply virtual compton scattering ( dvcs ) .
the main diagram for this reaction is shown in fig .
[ handbag ] . to there are 4 ( flavour sets of ) gpd s : 2 unpolarized functions ( @xmath61 and @xmath17 ) and 2 polarized ( @xmath62 and @xmath63 ) .
the two functions @xmath17 and @xmath63 correspond to helicity flip operators and have no direct analogon in the forward pdf s .
the functions @xmath61 and @xmath62 have as their forward limit the standard unpolarized and polarized pdf s @xmath22 and @xmath21 , respectively . the gpd s depend on 3 independent kinematic variables : @xmath64 where @xmath65 is related to the usual bjorken variable by @xmath66 .
( note that there are other equivalent choices possible for the variables . )
the importance of gpd s in the context of spin physics is embodied by the sum rule which was derived by ji @xcite for the second moment of the gpd s : @xmath67dx = j_q \label{ji}\ ] ] where @xmath68 is the _ total _ angular momentum contribution of quarks to the spin of the nucleon . up to now
there is experimental information only on the contribution from the spins of quarks ( and gluons ) . combined with this sum rule gpd s and
thus exclusive reactions are the only way presently known that may lead to information on the orbital angular momentum contribution , thus providing a complete picture of the nucleon spin structure .
the main problem in the experimental determination of the gpd s is the fact that they do not appear as simple factors in the expressions for cross sections , asymmetries etc .
they always appear in convolutions over the @xmath1 and @xmath65 variables , making a direct determination impossible . finding the best deconvolution procedure to lead from the experimental data to
the gpd is a challenge for the future . to do this a good understanding of the structure of
the gpd is essential to develop adequate models .
[ dvcs1 ] shows the missing mass ( squared ) spectrum for the dvcs reaction at hermes .
a clear peak at the ground state mass can be seen , thus establishing the observation of exclusive dvcs . to however ,
at the hermes kinematics with relatively low beam energy the main source of hard photons is not dvcs but the bethe - heitler process where the photon is radiated from either the incoming or the outgoing electron .
it is in principle impossible to disentangle these two processes .
on the other hand the presence of two processes will also lead to interference between them and this can be exploited to access the dvcs amplitude .
the interference terms can be projected out by measuring azimuthal asymmetries in single photon electroproduction .
it was shown @xcite that beam - spin ( @xmath69 ) , target - spin ( @xmath70 ) and beam - charge ( @xmath71 ) azimuthal asymmetries provide access to different combinations of the real and imaginary parts of the interfering amplitudes .
it should be noted that at present hera is the only place where beam charge asymmetries can be studied because of the availability of both electron and positron beams . in fig .
[ dvcs2 ] the beam - spin azimuthal asymmetry for a proton target is shown , while fig .
[ dvcs3 ] presents the results for the beam - charge asymmetry on the proton .
the @xmath72 moment of the former one is related to the imaginary part of the interference amplitude ; the @xmath73 moment of the latter is related to the real part of the interference amplitude . to
to it is obvious from the results presented in these figures that the dvcs - bh interference indeed dominates the distributions .
the extracted @xmath72 and @xmath73 moments are in encouraging agreement with theoretical calculations based on models of gpd s . in a next step the kinematic distributions of the various moments of the azimuthal asymmetries are extracted and to be compared to more detailed calculations . from fig .
[ dvcs1 ] it can be seen that the experimental resolution of the hermes spectrometer is insufficient to allow a clear separation of exclusive and fragmentation processes . at present
only the scattered electron and the real photon are detected in the forward spectrometer . to really establish exclusivity on
the event level would require the detection of the recoil target proton .
kinematics dictate that this proton recoils with low momentum at high angles relative to the beam line , outside the present acceptance of the hermes spectrometer .
a major project at hermes now is the construction of a recoil detector to detect and identify such recoil particles .
the detector will consist of three active detector parts . a silicon detector around the target cell inside the beam vacuum , a scintillating fibre tracker in a longitudinal magnetic field and a layer of scintillator strips with interspersed w - sheets as preshower material .
it is planned to install the recoil detector somewhere in 2004 or 2005 and to operate it for on the order of 2 years with an unpolarized hydrogen target and polarized electron and positron beams .
the higher target densities that are possible for unpolarized targets will enable hermes to obtain a large data sample on dvcs with unprecedented statistical and systematic accuracy for both beam - spin and beam - charge asymmetries .
an added advantage of the recoil detector is that the determination of the momentum of the recoil proton immediately yields an accurate measurement of the momentum transfer @xmath74 to the target .
this will improve the resolution in this important kinematic quantity ( see e.g. eq . [ ji ] ) by about an order of magnitude .
up to now we have only discussed the helicity structure of the nucleon .
in fact there are three different structure functions which describe the parton structure of the nucleon at leading twist .
these are the unpolarized structure function @xmath22 , the helicity structure function @xmath21 and a new quantity , transversity @xmath75 .
the related pdf is denoted @xmath76 and gives the probability to find a transversely polarized quark in a transversely polarized nucleon .
it is experimentally unknown , mainly because as a chiral - odd object it can not be measured through inclusive dis . in combination with another chiral - odd object ,
e.g. a fragmentation function , transversity becomes accessible . in semi - inclusive reactions
it leads to single - spin azimuthal asymmetries in e.g. pion and kaon production on a transversely polarized target .
such a transversely polarized target is installed in hermes since 2001 .
first data have been taken with it after hera became operational again following a major luminosity upgrade for the collider experiments . however , at the present time the collected data sample is too small to make a measurement of the transversity distributions feasible .
this will come in the run of 2003 and 2004 . on the other hand , hermes
has already collected a large amount of data on a longitudinally polarized target .
such a target is longitudinal with respect to the incoming electron beam , but does contain a small ( @xmath77 ) transverse component with respect to the virtual photon .
moreover , there is a number of higher - twist distribution functions and fragmentation functions which are directly related to the transversity distribution and which also cause single - spin asymmetries in semi - inclusive reactions on a longitudinal target .
the experimental single - spin azimuthal asymmetry for positive and negative pions on a longitudinally polarized proton target is depicted in fig .
the positive pions show a clear @xmath72 modulation of the asymmetry , while this is absent for negative pions .
this is consistent with the picture of @xmath45-quark dominance
. a larger set of results on such asymmetries for a deuteron target was recently published @xcite .
theoretical calculations based on models of the transversity distribution have had some success in reproducing these data . to
some recent results from the hermes experiment at hera were presented .
the analysis of the large data set on longitudinally polarized proton and deuteron targets , collected in the first run of hermes from 1995 through 2000 , is nearing completion .
results of a nlo qcd analysis of the inclusive scattering show indications for a positive and possibly large gluon polarization .
the results from the semi - inclusive analysis are consistent with those from the inclusive data and moreover give for the first time a flavour decomposition of the quark polarizations in 5 components .
the valence quarks are seen to dominate the contribution of the quark spins to the nucleon spin , while the sea appears to have a negligible polarization .
the observation of azimuthal asymmetries in exclusive reactions , here exemplified by the dvcs reaction , promises access to the new gpd s . the implementation of the recoil detector now under construction at hermes will give a much enhanced accuracy for such studies .
the main thrust of the data taking of hermes at the present time is with a transversely polarized proton target . on the basis of single - spin azimuthal asymmetries observed in semi - inclusive scattering off a longitudinally polarized target
it can be expected that these data will give first experimental information on the transversity distribution of quarks in the nucleon . | recent results from the hermes experiment at hera are described .
the large data set from run - i has yielded new information on the helicity structure of the nucleon .
the data to be taken in run - ii will deal mainly with the transverse spin structure and with exclusive reactions . |
the _ effective dimension _ and _ strong effective dimension _ of an infinite binary sequence @xmath4 are defined as @xmath6 where @xmath7 is the kolmogorov complexity of @xmath8 .
they can be characterized as effective versions of hausdorff and packing dimensions respectively , or by divergence of @xmath9-gales ( see @xcite for the original results and @xcite for a survey ) .
let @xmath10 $ ] be a computable real number and @xmath11 the bernoulli measure over cantor space given by @xmath12=p^{|w|_1}(1-p)^{|w|_0}$ ] .
it is well - known that if an infinite binary sequence @xmath4 is martin - lf random w.r.t .
@xmath11 then @xmath13 , where @xmath14 is the entropy of @xmath11 defined by @xmath15 this result is not difficult to prove and reduces to the strong law of large numbers for martin - lf random sequences , as on the one hand is the prefix version of kolmogorov complexity ] @xmath16+o(1)\ ] ] for @xmath11-random sequences by levin - schnorr theorem , and on the other hand @xmath17=-\frac{|x{\ensuremath{{\mathbin\upharpoonright}\raise-.2ex\hbox{\scriptsize $ n$}}}|_1}{n}\log(p)-\frac{|x{\ensuremath{{\mathbin\upharpoonright}\raise-.2ex\hbox{\scriptsize $ n$}}}|_0}{n}\log(1-p)\ ] ] which converge to @xmath14 for @xmath11-random sequences , by the strong law of large numbers for martin - lf random sequences .
this result highlights the relationship between shannon s information theory , kolmogorov algorithmic information theory and effective randomness .
ergodic theory provides a natural extension of information theory in which many results can be transferred , with more involved proofs , from the case of independent identically distributed random variables to the ergodic case , where independence is only required _ asymptotically , in the average _
( see section [ sec_back ] for a precise definition ) .
first , the strong law of large numbers extends to birkhoff s ergodic theorem .
second , the coincidence between local information and entropy extends through the shannon - mcmillan - breiman theorem .
whether martin - lf randomness fits with these theorems has been an open problem for a while .
the first results were proved by vyugin @xcite , based on non - classical , constructive proofs of the theorems .
he proved , in particular : [ thm_birkhoff_vyugin ] let @xmath0 be a computable shift - invariant ergodic measure over @xmath1 and @xmath18 be computable .
for every martin - lf @xmath0-random sequence @xmath4 , @xmath19 the entropy of an ergodic measure is defined as @xmath20\log \mu[w].\ ] ] observe that ( [ entropy_bernoulli ] ) and ( [ eq_entropy2 ] ) are consistent as they give the same quantity when @xmath0 is a bernoulli measure .
[ thm_sh1 ] let @xmath0 be a computable shift - invariant ergodic measure over @xmath1 . for every martin - lf @xmath0-random sequence @xmath4 , @xmath21=h(\mu).\ ] ] the question whether @xmath22 coincides with @xmath23 for every martin - lf @xmath0-random was left open by vyugin .
an alternative proof of theorem [ thm_sh1 ] approximating ergodic measures by markovian measures was later developed by nakamura @xcite , but also left the question open . in this paper
we provide a positive answer to this question .
a classical proof of the shannon - mcmillan - breiman theorem uses birkhoff s ergodic theorem , applied to some particular functions . the problem in making it
effective is that these functions are not computable in general .
recent works have been achieved to push the effective ergodic theorem to the largest possible class of functions .
here we extend it enough to get the full effective shannon - mcmillan - breiman theorem .
in section [ sec_back ] we recall basic notions of computability , randomness and ergodic theory . in section [ sec_erg ]
we develop effective versions of birkhoff s ergodic theorem . in section [ sec_sh ]
we present our main result .
we work on the cantor space @xmath24 of infinite binary sequences . a finite word @xmath25 determines the cylinder @xmath26\subseteq{\{0,1\}^{\mathbb{n}}}$ ] of infinite sequences starting with @xmath8 . if @xmath27 and @xmath28 , @xmath29 is the prefix of @xmath4 of length @xmath30 , and is also denoted @xmath31 .
the cylinders form a base of the product topology .
[ [ effective - topology . ] ] effective topology .
+ + + + + + + + + + + + + + + + + + + an open set @xmath32 is _ * effective * _ if it is a recursively enumerable union of cylinders .
a closed set is effective it its complement is an effective open set .
a function @xmath33 is _ * computable * _ if there is a turing machine that on oracle @xmath4 and input @xmath30 computes a rational number @xmath34 such that @xmath35 .
equivalently , @xmath36 is computable if for every rational numbers @xmath37 , @xmath38 is effectively open , uniformly in @xmath39 .
a function @xmath40 $ ] is _ * lower * _ ( resp . _ * upper * _ ) _ * semi - computable * _ if there is a turing machine that on oracle @xmath4 and input @xmath30 computes a rational number @xmath41 such that @xmath42 ( resp .
@xmath43 ) .
equivalently , @xmath36 is lower ( resp .
upper ) semi - computable if for every rational number @xmath44 , @xmath45 $ ] ( resp .
@xmath46 ) is effectively open , uniformly in @xmath44 .
[ [ kolmogorov - complexity - and - martin - lf - randomness . ] ] kolmogorov complexity and martin - lf randomness .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + for @xmath25 , @xmath7 is the prefix version of kolmogorov complexity , defined by levin and chaitin independently .
it is defined as the length of a shortest input of a universal turing machine with prefix - free domain computing @xmath8 on that input .
a probability measure @xmath0 over @xmath24 is determined by its value on cylinders @xmath47 $ ] , for @xmath25 .
@xmath0 is computable if @xmath47 $ ] is a computable real number , uniformly in @xmath8 .
given a computable probability measure @xmath0 , a sequence @xmath27 is _ * martin - lf @xmath48-random * _ , denoted @xmath49 , if there is @xmath50 such that for all @xmath30 , @xmath51-c.\ ] ] martin - lf s original definition @xcite was expressed in terms of tests rather than complexity , but the one given here , due to levin and chaitin @xcite independentely , was proved to be equivalent by levin @xcite and schnorr @xcite .
the function @xmath52-k(x{\ensuremath{{\mathbin\upharpoonright}\raise-.2ex\hbox{\scriptsize $ n$}}})\}\ ] ] is lower semi - computable and @xmath53 .
moreover , it was proved in @xcite that @xmath54 is maximal in the sense that for every integrable lower semi - computable function @xmath40 $ ] , there exists @xmath55 such that @xmath56 .
we call such an @xmath36 a _ * @xmath48-test*_. it tests martin - lf randomness in the sense that @xmath49 iff @xmath57 for each @xmath0-test @xmath36 iff @xmath58 .
more can be found on this subject in @xcite .
[ [ ergodic - theory . ] ] ergodic theory .
+ + + + + + + + + + + + + + + we recall some basic notions of ergodic theory , more details can be found in @xcite .
we denote by @xmath59 the _ * shift map * _ defined by @xmath60 .
a measure @xmath0 over @xmath24 is _ * shift - invariant * _ if for all borel sets @xmath61 , @xmath62 , equivalently if @xmath63+\mu[1w]=\mu[w]$ ] for all @xmath25 .
@xmath0 is _ * ergodic * _ if for all borel sets @xmath61 such that @xmath64 up to a null sets , @xmath65 or @xmath66 .
equivalently , @xmath0 is ergodic if for all @xmath67 , @xmath68\cap t^{-k}[v])=\mu[u]\cdot \mu[v].\ ] ]
the following theorem , taken from @xcite , extends a result of ku@xmath69era from the uniform measure to any ergodic shift - invariant measure : [ thm_poincare ] let @xmath0 be a computable ergodic shift - invariant measure and @xmath70 an effective closed set such that @xmath71 .
every martin - lf @xmath0-random sequence has a tail in @xmath72 , i.e. for every @xmath49 there exists @xmath73 such that @xmath74 . in @xcite and @xcite independently this result was used to prove that not only the orbit of @xmath4 eventually falls into @xmath72 , but it does so with frequency @xmath75 .
[ thm_brk_lsc ] let @xmath0 be a computable ergodic shift - invariant measure and @xmath70 an effective closed set such that @xmath71 .
for every martin - lf @xmath0-random sequence @xmath4 , @xmath76 we first generalize the result from sets to functions : [ thm_erg_semi ] let @xmath0 be a computable ergodic shift - invariant measure .
assume @xmath77 $ ] is : * either lower semi - computable , * or upper semi - computable and bounded by a @xmath0-test . for each @xmath49 , @xmath78 let us introduce the notation @xmath79 .
if @xmath36 is lower semi - computable , then there is a sequence of uniformly computable nonnegative functions @xmath80 .
applying theorem [ thm_birkhoff_vyugin ] to @xmath81 and @xmath49 gives @xmath82 . by the monotone convergence theorem , @xmath83 , so @xmath84 . if @xmath85 we are done .
otherwise , let @xmath86 be a rational number .
the set @xmath87 is effectively closed and by the classical ergodic theorem , there exists @xmath88 such that @xmath89 .
theorem [ thm_poincare ] tells us that if @xmath49 then there is @xmath30 such that @xmath90 . as a result , @xmath91 . as
this is true of every @xmath86 , we get the result . now ,
if @xmath36 is upper semi - computable and @xmath92 where @xmath93 is a @xmath0-test , then for @xmath49 , applying the preceding result to @xmath93 and @xmath94 , @xmath95 we then extend this result further : [ cor_delta2_bounded ] let @xmath40 $ ] be @xmath96 on @xmath97 , i.e. there is a sequence @xmath81 of uniformly computable functions such that @xmath98 for each @xmath49 .
assume that @xmath36 is dominated by a @xmath0-test .
for every @xmath49 , @xmath99 let @xmath100 and @xmath101 .
on @xmath97 , @xmath102 and @xmath103 . by the monotone and dominated convergence theorem , the convergences hold in @xmath104 . applying theorem [ thm_brk_lsc ] to @xmath105 and @xmath106
gives the result .
more precisely , for every @xmath49 and every @xmath107 , @xmath108 so @xmath109
we now present our main result . a proof of the classical result , stating the result for a.e .
@xmath4 , can be found in @xcite .
it makes use of martingale convergence theorems and ergodic theorems . the main difficulty in adapting the proof is to make sure that the effective versions of the ergodic theorem can be applied .
the rest of this section is devoted to the proof of theorem [ thm_main ] .
an easy calculation shows that @xmath111 = \sum_{k=0}^{n-1}f_{n-1-k}\circ t^{k}(x)\ ] ] where @xmath112=-\log\frac{\mu[x_0\ldots x_k]}{\mu[x_1\ldots x_k]}&\quad & \text{for } k\geq 1,\\ f_0(x ) & : = -\log \mu[x_0].\end{aligned}\ ] ] define the computable martingale @xmath114 } \\
d(x_0\ldots x_k ) & = \frac{\mu[x_1\ldots x_k]}{\mu[x_0\ldots x_k]}\quad\text{for $ k\geq 1$.}\end{aligned}\ ] ] by the effective doob s convergence theorem ( see theorem 7.1.3 on page 270 in @xcite ) , for each @xmath49 , @xmath115 converges , and so does @xmath116 .
let @xmath117 be the limit .
we write @xmath118=\frac{1}{n}\sum_{k=0}^{n-1}(f_{n-1-k}\circ t^k(x)-f\circ t^k(x))+\frac{1}{n}\sum_{k=0}^{n-1}f\circ t^k(x)\ ] ] and prove that the first term tends to @xmath119 while the second term converges to @xmath120 .
@xmath128 is lower semi - computable and by lemma [ lem_sup ] it is a @xmath0-test . by construction
, @xmath36 is @xmath96 on @xmath97 and it is dominated by @xmath128 so it satisfies the conditions of corollary [ cor_delta2_bounded ] , from which the result follows directly .
let @xmath130 for @xmath131 , @xmath132 so @xmath133 .
as @xmath122 a.e . , @xmath134 a.e .
as @xmath135 , @xmath134 in @xmath124 by the dominated convergence theorem . on @xmath97 , @xmath136 fix @xmath107 and
let @xmath137 .
as @xmath138 is lower semi - computable , the first term converges to @xmath139 by the effective ergodic theorem [ thm_erg_semi ] .
as @xmath140 is @xmath96 on @xmath97 and is dominated by the @xmath0-test @xmath141 , the second and the third terms converge to @xmath142 by corollary [ cor_delta2_bounded ] so their limits cancel each other .
laurent bienvenu , adam day , ilya mezhirov , and alexander shen .
ergodic - type characterizations of algorithmic randomness . in _
computability in europe ( cie 2010 ) _ , volume 6158 of _ lecture notes in computer science _ , pages 4958 .
springer , 2010 .
johanna n.y .
franklin , noam greenberg , joseph s. miller , and keng meng ng .
artin - lf random points satisfy birkhoff s ergodic theorem for effectively closed sets . to appear in the _ proceedings of the american mathematical society _ , 2010 . | let @xmath0 be a computable ergodic shift - invariant measure over @xmath1 . providing a constructive proof of shannon - mcmillan - breiman theorem
, vyugin proved that if @xmath2 is martin - lf random w.r.t .
@xmath0 then the strong effective dimension @xmath3 of @xmath4 equals the entropy of @xmath0 .
whether its effective dimension @xmath5 also equals the entropy was left as an problem question . in this paper
we settle this problem , providing a positive answer . a key step in the proof
consists in extending recent results on birkhoff s ergodic theorem for martin - lf random sequences .
+ _ keywords : shannon - mcmillan - breiman theorem ; martin - lf random sequence ; effective hausdorff dimension ; compression rate ; entropy . |
in pure extreme type - ii superconductors , such as the high @xmath0 superconductors , the abrikosov vortex line lattice melts via a sharp first order phase transition@xcite into a vortex line liquid as the temperature is increased above a critical @xmath1 .
the properties of this vortex line liquid phase have been the subject of considerable investigation .
theoretical arguments@xcite and early simulations@xcite suggested that the vortex line liquid might retain superconducting phase coherence parallel to the applied magnetic field , within some temperature interval above @xmath1 .
later , better converged simulations@xcite found that phase coherence is simultaneously lost in all directions upon melting .
subsequently , teanovi@xcite has proposed that , for small magnetic fields , there still remains a sharp thermodynamic phase transition at a temperature @xmath2 within the vortex liquid state , associated with diverging fluctuations of closed vortex loops , such as drive the superconducting transition in the zero magnetic field case .
considering the limit of infinite penetration length @xmath3 , teanovi proposed that , in a finite field , the fluctuations of the magnetic field induced vortex lines act to screen the interactions of thermally excited closed vortex loops , in the same way that magnetic field fluctuations screen the vortex loop interactions of a finite @xmath3 model in zero applied magnetic field . pursuing this argument
, he predicted that the proposed vortex loop blowout transition at @xmath2 may be an inverted xy transition , as is the case of the zero field meissner transition for the finite @xmath3 model .
following teanovi s predictions , sudb and co - workers@xcite have carried out numerical simulations of the three dimensional ( 3d ) uniformly frustrated xy model of a type - ii superconductor .
they claim to find evidence for teanovi s transition , which they associate with the formation of a vortex line path that percolates entirely around the system in the direction transverse to the magnetic field .
most recently , measurements@xcite on high purity yba@xmath4cu@xmath5o@xmath6 ( ybco ) single crystals have found a step like anomaly in the specific heat at a temperature highter than the melting @xmath1 , reminiscent of an inverted mean field transition .
it has been argued@xcite that this feature may be evidence for teanovi s transition @xmath2 . in order to further investigate this issue ,
we have carried out new simulations on the 3d uniformly frustrated xy model , both repeating the approach of sudb and co - workers , and measuring new quantities that make a more direct test of teanovi s theory . after correcting certain inconsistencies in the earlier numerical work ,
we show that whether or not one finds indications of a vortex loop blowout transition depends crucially on how one chooses to resolve vortex line paths at points where two or more lines intersect . making the choice that favors the blowout interpretation
, we find the critical exponent @xmath7 , rather than the value @xmath8 expected for an inverted 3d xy transition .
finally , we make high precision measurements of the specific heat , in search of a thermodynamic signature for a blowout transition , but no such signature is found .
the model that we use is the 3d uniformly frustrated xy model@xcite which models a type - ii superconductor in the limit of infinite magnetic penetration length , @xmath9 , and is given by the hamiltonian , @xmath10 = -\sum_{i , \mu}j_{\mu}\cos ( \theta_{i+\hat\mu}-\theta_{i}-a_{i\mu})\enspace . \label{eqh}\ ] ] here @xmath11 are the nodes of a cubic grid of sites , @xmath12 @xmath13 , are the directions of the grid axes , and the sum is over all nearest neighbor bonds of the grid . @xmath14 is the phase angle of the superconducting wavefunction on site @xmath11 , @xmath15 is given by the integral of the magnetic vector potential @xmath16 across the bond at site @xmath11 in direction @xmath17 , and @xmath18 is the flux quantum .
the argument of the cosine is the gauge invariant phase angle difference across the bond .
the circulation of the @xmath19 around any plaquette of the grid is equal to @xmath20 times number of flux quanta of magnetic field penetrating the plaquette .
we take the magnetic field , @xmath21 , uniform and parallel to the @xmath22 axis , with a fixed density of flux quanta @xmath23 per plaquette of area @xmath24 . the couplings @xmath25 we take to model an anisotropic system , with @xmath26 , and @xmath27 .
simulations are carried out varying the @xmath28 according to a usual monte carlo scheme ; the @xmath19 are held fixed . that the @xmath19 do not fluctuate , and that they give a uniform magnetic field , are the consequences of the @xmath9 approximation .
simulations are carried out on @xmath29 cubic grids , using periodic boundary conditions .
except where otherwise noted , our runs are typically for @xmath30 to @xmath31 monte carlo passes through the entire lattice . while we simulate in the phase angle degrees of freedom , @xmath28 , our interest will be in the behavior of the vorticity in these phase angles .
let @xmath32 denote the _ dual _ sites of the original grid ; these are the sites at the centers of the unit cells of the grid .
denote by @xmath33 the plaquette which is the face of the unit cell centered on dual site @xmath32 with normal in the @xmath17 direction , @xmath34 , @xmath13 .
we define the integer vorticity @xmath35 piercing plaquette @xmath33 by computing the circulation of the gauge invariant phase angle differences around the plaquette , @xmath36 = 2\pi ( n_{s\mu } -f\delta_{z\mu})\enspace,\ ] ] where the sum is counterclockwise around all bonds forming the boundary of plaquette @xmath33 , and the gauge invariant phase angle differences are restricted to the interval @xmath37 . in a constant magnetic field
, the condition that the total energy density remains finite can be shown to yield the `` neutrality '' constraint ( see section iii.a ) , @xmath38 i.e. the total vorticity piercing any plane at constant @xmath13 is @xmath39 ; these are the magnetic field induced vortex lines .
the total vorticity in the transverse directions @xmath40 and @xmath41 is zero . taking the vorticity @xmath35 as the directed bond of the dual grid , emanating from site @xmath32 in direction @xmath17 , the vorticity so defined is divergenceless , forming continuous lines that , due to the periodic boundary conditions , must ultimately close upon themselves .
we will label such a closed vortex path by the index @xmath42 , and define the vector @xmath43 as the net displacement one travels upon following the path @xmath42 from a given starting point until returning back to that point as the line closes back on itself . if @xmath44 , then the vortex line path is a closed loop of finite extent that exists as a thermal fluctuation .
if @xmath45 , @xmath46 integer , then such a vortex line path represents @xmath46 of the @xmath39 field induced vortex lines ; these @xmath46 lines are mutually connected to each other via the periodic boundary conditions in the @xmath13 direction.@xcite for @xmath47 , we can say that the @xmath46 field induced lines are geometrically entangled with each other .
if @xmath48 or @xmath49 , then the vortex line path winds @xmath46 times around the system _ transversely _ to the applied magnetic field
. we will be particularly interested in vortex line paths for which @xmath50 , but @xmath51 or @xmath52 .
henceforth , we will refer to as the `` lines '' , the set of vortex line paths @xmath53 for which all @xmath54 ; these are the field induced vortex lines .
all other vortex paths we will refer to as the `` loops '' . in order to trace out vortex line paths , one needs to know how to treat intersections .
an intersection is when there is more than one vortex line entering and exiting a give unit cell of the grid , and it is therefore ambiguous which entering segment to connect to which exiting segment .
it was previously shown@xcite by one of us that the method chosen to resolve such intersections can have a dramatic effect on the statistics of closed thermally excited loops in the zero field @xmath55 model . here ,
for the @xmath56 model , we consider two different schemes , which we henceforth refer to as method ( i ) and method ( ii ) : \(i ) at each intersection we choose randomly , with equal probability , which entering segment connects to which exiting segment . in the @xmath55 model
this scheme was found to give results closest to theoretical expectations.@xcite \(ii ) motivated by sudb and co - workers,@xcite we first search@xcite through all possible connections to find a path @xmath42 with @xmath50 and @xmath51 or @xmath52 .
such a path winds around the system transverse to the field , without ever winding around the system parallel to the field .
if one such path is found , it is selected as a path @xmath57 contributing to the `` loops '' , and we then repeat the proceedure applied to all remaining vortex paths .
when all such transverse paths are found , the remaining vortex line intersections are resolved randomly , as in method ( i ) . using either method
( i ) or method ( ii ) we thus decompose the vorticity of any given configuration into disjoint closed vortex line paths , consisting of a set @xmath53 of `` lines '' and a set @xmath58 of `` loops '' .
we first attempt a direct test of teanovi s theory of the @xmath2 transition within the liquid phase .
a summary of his arguments for the existence of this transition is as follows .
first , a duality transformation@xcite from the xy model of eq.([eqh ] ) gives the interaction between vortices as , @xmath59={1\over 2}\sum_{s , s^{\prime},\mu } [ n_{s\mu}-f\delta_{z\mu}]v^{\mu}({\bf r}_{s}-{\bf r}_{s^{\prime}})[n_{s^{\prime}\mu}-f\delta_{z\mu } ] \label{eqv}\ ] ] where @xmath60 is the appropriate anisotropic generalization of the coulomb interaction , with fourier transform @xmath61 .
it is this singularity of @xmath62 as @xmath63 that yields the constraint of eq.([eqneutral ] ) .
next , one imagines decomposing the total vorticity of the system into lines and loops , @xmath64 if we define , @xmath65 then @xmath66 and the hamiltonian of eq.([eqv ] ) can be rewritten as , @xmath67v^{\mu}({\bf r}_{s}-{\bf r}_{s^{\prime}})[n_{s^{\prime}\mu}^\mathrm{loops}-b_{s^{\prime}\mu}]\enspace . \label{eqloop}\ ] ] teanovi then argues that a coarse graining of vortex fluctuations , in the vortex line liquid phase , leads to an effective hydrodynamic hamiltonian on long length scales which has the same interaction piece as eq.([eqloop ] ) , but which has a new additive term proportional to @xmath68 .
the resulting long length scale hamiltonian then has exactly the same form as that of a zero field superconductor with thermally fluctuating vortex loops , @xmath69 , _ and _ a thermally fluctuating magnetic field @xmath70 whose average is zero , i.e. the zero field superconductor with a finite penetration length @xmath3 . in other words , in this _ infinite _ @xmath3 theory at _ finite _ magnetic field , the long wave length fluctuations of the field induced vortex lines @xmath71 screen the interaction between the vortex loops @xmath69 in exactly the same manner as magnetic field fluctuations screen the interactions between vortex loops in a _
@xmath3 model at _ zero _ magnetic field . the meissner transition at @xmath72 in the zero field , finite @xmath3 , model is an _ inverted _
3d xy transition.@xcite the high temperature phase @xmath73 has vortex loops on all length scales and breaks a global @xmath74 symmetry associated with a _ disorder _
parameter;@xcite the low temperature phase @xmath75 has no vortex loops on sufficiently long length scales .
the correlation length @xmath76 and renormalized magnetic penetration length @xmath77 both diverge@xcite as @xmath78 .
we have earlier carried out numerical simulations@xcite of this zero field , finite @xmath3 , meissner transition .
we demonstrated that , in this model , magnetic field fluctuations obey the finite size scaling relation , @xmath79 where in the above @xmath80 and @xmath81 is the fourier transform of of the magnetic flux density @xmath70 .
as @xmath82 , and @xmath63 , @xmath83 hence @xmath84 vanishes below the transition , and increases continuously from zero as one goes above the transition . in the present case of a finite magnetic field , if teanovi s mapping is correct , the meissner transition @xmath72 becomes the transition @xmath85 within the vortex line liquid phase , and we expect the exact same scaling as in eq.([eqscal ] ) above , when applied to the quantity @xmath70 defined in eq.([eqb ] ) .
taking the limit of @xmath63 in eq.([eqscal ] ) , and applying to systems with fixed aspect ratio @xmath86 , we expect the scaling , @xmath87 where @xmath88 and @xmath89 . for the directions @xmath90 or @xmath91
, @xmath92 is the net vorticity of the magnetic field induced vortex lines in the transverse direction @xmath17 .
the two dimensional vector @xmath93 defined above is the integer valued `` winding number '' that counts the net number of times the field induced vortex lines wind around the system in the transverse directions @xmath94 and @xmath91 . if @xmath53 is the set of vortex line paths that define the field induced vortex lines @xmath71 , and @xmath95 is the net displacement along path @xmath42 as defined earlier , then @xmath96 , where @xmath97 .
we thus expect from eq.([eqscal2 ] ) the finite size scaling , @xmath98 note , the neutrality condition of eq.([eqneutral ] ) implies that the _ total _ transverse vorticity in the system must always vanish . for @xmath99 , it is therefore necessary that any such winding of the field induced lines is exactly canceled out by an equal and opposite transverse winding of the loops . in the thermodynamic limit , @xmath100 , eq.([eqscal1 ] )
implies that @xmath101 for @xmath102 , and @xmath103 increases continuously from zero as one increases @xmath104 . the proposed transition at @xmath85 is thus associated with the appearance of infinite transverse loops ( see following section iv ) .
another interpretation of the @xmath85 transition follows from the `` two dimensional ( 2d ) boson '' model@xcite of interacting vortex lines , in which the field induced vortex lines are viewed as the world lines of two dimensional bosons traveling down the imaginary time axis . for @xmath102 where @xmath105 , the field induced vortex lines behave like _ charged _ two dimensional bosons,@xcite with a long range retarded coulomb interaction . in the vortex line liquid , @xmath106 , where phase coherence is lost parallel to the applied magnetic field ,
the analog 2d bosons are in a _ charged _ superfluid state . for @xmath104 , where @xmath107 , screening by the infinitely large loops @xmath69 results in an effective short ranged interaction between the field induced lines . in this case
the winding number squared @xmath108 is proportional to the superfluid density of what is now an _
uncharged _ superfluid .
thus @xmath85 corresponds to a transition between a charged superfluid and an uncharged superfluid in the analog 2d boson theory .
equivalently , if one considers the quanta that mediate the interaction between the analog 2d bosons , the transition is from massless quanta for @xmath102 to massive quanta for @xmath104 . to arrive at eq.([eqscal3 ] ) , we considered the transverse components of eq.([eqscal2 ] ) .
however , we can also consider the parallel , @xmath109 component .
now , @xmath110 and so we expect the scaling @xmath111 if @xmath53 is the set of vortex line paths that define the field induced vortex lines @xmath71 , and @xmath95 is the net displacement along path @xmath42 as defined earlier , then @xmath112 .
thus @xmath113 gives the number of `` lines '' in excess of the average value @xmath39 set by the applied magnetic field .
the neutrality condition of eq.([eqneutral ] ) requires that when @xmath114 , there must be an equal and opposite parallel winding of the loops @xmath69 . as @xmath82 , we have @xmath115 for @xmath102 , and @xmath114 for @xmath104 .
thus a transition at @xmath85 should be characterized by fluctuations in the number of field induced lines and by the appearance of infinite parallel loops directed opposite to the direction of the applied magnetic field . to test the above predictions , we have simulated the 3d uniformly frustrated xy model of eq.([eqh ] ) using a vortex density @xmath116 , anisotropy @xmath117 , and aspect ratio @xmath118 , for @xmath119 @xmath120 , @xmath121 , @xmath122 and @xmath123 . for these parameters ,
the vortex lattice melting temperature is @xmath124 , and the zero field critical temperature is @xmath125 .
we compute the transverse winding @xmath126 of the field induced lines , defined by eq.([eqwp ] ) , using both method ( i ) and method ( ii ) to decompose each configuration into `` lines '' and `` loops '' .
according to the scaling equation ( [ eqscal3 ] ) , we expect that plots of @xmath108 vs. @xmath127 for different sizes @xmath128 should all intersect at the common point @xmath129 , or @xmath130 . in fig.[f1 ]
we show a semilog plot of @xmath108 vs. @xmath131 using method ( i ) ( random reconnections at intersections ) for @xmath132 , @xmath120 , and @xmath121 .
we see that there is clearly no common intersection point of the curves . as @xmath128 increases ,
@xmath108 decreases uniformly over the entire temperature range .
this is in qualitative agreement with earlier computations of @xmath108 by one of us ( see fig . 15 of ref . ) . for @xmath133 ,
we have found no net transverse winding of the field induced lines at all , i.e. for the length of our simulation we had @xmath134 , for the temperature range @xmath135 .
= 3.2truein next , in fig.[f2 ] ,
we show the same quantities but now using method ( ii ) ( search first for maximal transverse loops ) , for @xmath119 @xmath120 , @xmath121 , @xmath122 and @xmath123 .
we see that as @xmath128 increases , the curves do seem to approach a common intersection point , giving a @xmath136 . note that this @xmath85 is _ above _ the zero field critical temperature @xmath125 . from eq.([eqscal3 ] ) , we expect that the slopes of these curves at @xmath85 should scale with system size as , @xmath137 .
fitting each of the curves of @xmath138 to a cubic polynomial in @xmath127 , we compute their derivatives at the intersection point @xmath139 , and plot the results vs. @xmath140 in fig.[f2.1 ] .
we see that the slopes , to an excellent approximation , scale linearly with @xmath140 , thus suggesting a critical exponent @xmath7 . on closer inspection ,
the data in fig.[f2.1 ] show a small systematic downwards curvature about the linear fit ; however this curvature can be removed by assuming a slightly higher critical temperature of @xmath141 .
note that this value of @xmath7 is _ larger _ than the predicted value of @xmath142 . as an alternative method to compute the critical behavior
, we can take the scaling equation ( [ eqscal3 ] ) , expand the scaling function @xmath143 as a polynomial for small @xmath40 , and do a nonlinear fitting to the data to determine the unknown polynomial coefficients , @xmath85 , and @xmath144 . to obtain the best fit we use a 4th order polynomial and fit only the data from the two largest sizes , @xmath145 and @xmath123 .
the results give @xmath146 and @xmath147 , in agreement with the earlier estimates . in fig.[f3 ] we show the scaling collapse that results from this polynomial fit .
there are systematic deviations from the fitted curve on the @xmath104 side , though these appear to decrease as @xmath128 increases .
= 3.2truein = 3.2truein = 3.2truein next we consider the excess parallel winding of the field induced lines @xmath148 . as discussed earlier ,
teanovi s theory predicts a scaling of @xmath149 as in eq.([eqscalz ] ) . to determine @xmath148 we count the winding of vortex line paths @xmath53 that wind negatively in the @xmath13 direction , i.e. have a net displacement of @xmath150 , with @xmath151 a positive integer ( @xmath152 may have any value for such paths ) .
since such negative parallel windings must be compensated for by excess field `` lines '' , we have @xmath153 .
however , when we have used either of our tracing methods ( i ) or ( ii ) , we have _ never _ found _ any _ such negative parallel windings up to the highest temperature we have simulated , @xmath154 .
this has motivated us to define a third tracing scheme : ( iii ) we first search through all possible connections to find any paths with @xmath155 . in fig.[f3.1 ]
we show results , using tracing scheme ( iii ) , for @xmath149 vs. @xmath156 for the same system parameters and sizes as used in fig.[f2 ] for @xmath157 .
note that the values of @xmath149 where the curves for different @xmath140 intersect are exceedingly small .
the intersection points appear to decrease in @xmath127 as @xmath140 increases , however we do not have the accuracy to make any firm conclusions . in an attempt to improve the analysis of @xmath149 we have repeated the calculation , but using a new system aspect ratio of @xmath158 .
this has the effect of increasing the value where the curves of @xmath149 intersect , and so hopefully improving our accuracy .
we have explicitly checked that changing the aspect ratio does not shift the transition temperature @xmath85 that is observed in @xmath157 ( see also section iv.b ) . in fig.[f3.2 ] we show results for @xmath149 vs. @xmath156 for this new aspect ratio .
again we find no common intersection point for the sizes considered .
as @xmath140 increases , the intersection point continues to decrease . whether this is a failure of the scaling hypothesis of eq.([eqscalz ] ) , or whether we have simply failed to reach the scaling limit of sufficiently large @xmath159 ( @xmath160 is the largest value in fig.[f3.2 ] ) , we can not be certain .
note that in both figs.[f3.1 ] and [ f3.2 ] , @xmath149 appears to be vanishing at a temperature noticeably _ above _ the @xmath161 where the curves of the transverse winding , @xmath157 , intersect .
we have also tried to fit the data of fig.[f3.2 ] to the scaling form , @xmath162 , assuming a non trivial anomalous scaling dimension @xmath40 ( although we have no specific theoretical reason to propose this form ) .
when we do so , we obtain @xmath163 , @xmath164 , and @xmath165 , however our data in the vicinity of this @xmath0 is too scattered for us to place much significance on this fit .
having used tracing method ( iii ) to first eliminate all possible lines percolating in the negative @xmath22 direction , we can then go and search for all possible transversely percolating lines and compute the resulting transverse winding @xmath157 .
when we do this , we find our results for @xmath157 virtually unchanged from tracing method ( ii ) in the vicinity of @xmath161 .
the extremely low number of negative @xmath22 percolating lines at this temperature produces no noticeable effect on the transverse tracing .
= 3.2truein = 3.2truein
as discussed in the preceding section iii.a , a transition at @xmath85 would mark the appearance of infinite transverse loops , as @xmath127 is increased . the idea to explicitly look for transverse paths that percolate across the system
was first put forth by jagla and balseiro.@xcite later , sudb and co - workers@xcite refined this idea .
they defined a quantity which they denoted @xmath166 , which is the probability that a vortex path exists which travels completely across the system in a direction transverse to the applied magnetic field , without ever traveling completely across the system in the direction parallel to the field .
if such a path exists in a given configuration , that configuration counts as unity in the average for @xmath166 ; if not , that configuration counts as zero . since having @xmath167 in a given configuration necessarily implies that there is a percolating transverse loop in that configuration , there is a close connection between the quantities @xmath108 and @xmath166 .
they differ in that ( i ) for a configuration with @xmath168 , and hence with more than one percolating transverse loop , the contribution to @xmath166 remains unity , rather than increasing with the number of percolating transverse loops ; and ( ii ) in a configuration with two percolating but oppositely oriented transverse loops , the contribution to @xmath166 will be unity , but these loops cancel each other in their contribution to @xmath169 , which might therefore be zero . since @xmath166 is a pure number one might expect it to be a scale invariant quantity , and hence , similar to @xmath108 , plots of @xmath166 vs. @xmath127 for different system sizes @xmath128 should have a common intersection point at @xmath85 . sudb and co - workers method of searching for such percolating transverse paths is similar to our method ( ii ) except for one crucial difference.@xcite they do not require that the transverse path closes upon itself ; they only require that the path start at one end of the system , say at @xmath170 , and continue until reaching the opposite end , @xmath171 , while keeping the distance traveled along @xmath22 less than @xmath172 , that is the displacement traveled along the path @xmath42 satisfies @xmath173 and @xmath174 .
since , by the periodic boundary conditions , all paths _ must _ eventually close upon themselves , there are two possibilities for the transverse percolating paths that sudb and co - workers find .
we illustrate these in fig.[f4 ] : ( 1 ) the path @xmath42 eventually closes upon itself without ever traveling the length @xmath172 , in which case @xmath50 ; or ( 2 ) the path @xmath42 , when followed until it closes upon itself , does wind up traveling the length @xmath172 , with a displacement @xmath175 ; in this case , our method ( ii ) would consider this path as part of the field induced vortex lines @xmath71 , contributing to the winding @xmath126 , rather than as a transverse loop that contributes to @xmath69 .
we will call sudb s path tracing method ( ii@xmath176 ) to distinguish it from our method ( ii ) .
the probability for a percolating path using method ( ii@xmath176 ) we will denote by @xmath177 ; using method ( ii ) we will denote it by @xmath166 .
paths of type ( 2 ) will contribute to @xmath177 , but not to @xmath166 .
we will see that there are very dramatic differences between these two methods , and that only @xmath166 gives self - consistent results .
= 3.2truein first , we note that if we use method ( i ) ( random connections ) to search for percolating transverse paths , the result is essentially the same as found for @xmath157 in section iii.b . as @xmath128 increases , the probability to find a percolating transverse loop steadily decreases for the entire temperature range , becoming immeasurably small for our biggest system size .
hence we will focus here on methods ( ii ) and ( ii@xmath176 ) .
we now consider the computation of @xmath177 using method ( ii@xmath176 ) , the one used by sudb and co - workers , which never checks to see how the percolating transverse path closes upon itself .
we first use parameters @xmath178 and @xmath179 , the same as in section iii.b , but a more dilute density of vortex lines @xmath180 . for these parameters ,
the vortex lattice melting temperature is @xmath181 , and the zero field critical temperature , as before , is @xmath125 .
these parameters are very close to the parameters of one of the cases studied by sudb and co - workers in refs . and
( they used @xmath180 , @xmath182 and @xmath183 ) .
our results for @xmath177 vs @xmath156 , for three system sizes , @xmath184 , @xmath123 and @xmath185 , are shown as the solid symbols on the left hand side of fig.[f5 ] .
these results agree quite closely with those of sudb and co - workers ( see fig.8 of ref . ) , and seem to show what might be a common intersection of the three curves near @xmath186 .
however , we now consider the same parameters and sizes @xmath128 , only using a different system aspect ratio , @xmath187 .
the results are shown as the open symbols on the right hand side of fig.[f5 ] .
we see that there no longer appears to be a common intersection point , but more importantly , the curves have all shifted dramatically to higher temperatures .
thus , any value of @xmath85 that one might try to extract from @xmath177 depends sensitively on the system aspect ratio .
we have also considered other values of the aspect ratio @xmath188 , not shown here .
the clear trend is that the sharp rise in @xmath177 shifts to increasing temperatures as @xmath188 decreases .
but if @xmath85 represents a true phase transition , it must be _ independent _ of aspect ratio .
we therefore conclude that @xmath177 and method ( ii@xmath176 ) do not give any self - consistent evidence of the proposed vortex loop blowout transition .
= 3.2truein the problems with @xmath177 are even clearer if we consider the parameters @xmath116 and @xmath178 , the same ones used for our computation of @xmath108 in section iii.b .
in fig.[f6 ] we show our results for @xmath132 , @xmath120 , and @xmath122 , for the two aspect ratios @xmath179 and @xmath189 . in both cases ,
there is no common intersection point of the three curves for the three sizes , and the curves for the smaller aspect ratio are shifted to higher temperatures .
note also that , for both aspect ratios , the temperatures at which @xmath177 rises to unity lie quite significantly below the value of @xmath136 found from our analysis of @xmath108 .
= 3.2truein we next consider the the computation of @xmath166 using method ( ii ) ( percolating transverse path must close upon itself keeping @xmath50 ) . in fig.[f7 ] we show results using the same parameters as were used to compute @xmath177 in fig.[f5 ] , i.e. @xmath180 , @xmath178 , and @xmath184 , @xmath123 and @xmath185 , for the same two aspect ratios @xmath179 and @xmath187 .
we see now that for both aspect ratios , curves for the three different sizes appear to approach a common intersection point , @xmath190 , and that this intersection point is independent of aspect ratio ( note : for @xmath187 , the thinness of the system @xmath191 , for @xmath184 , presumably makes it too small to be in the scaling region , hence it intersects the other two curves at somewhat lower temperatures ) .
= 3.2truein in fig.[f8 ] we show similar results using the same parameters as were used in our computation of @xmath108 in section iii.b , i.e. @xmath116 , @xmath178 and @xmath179 .
we see that the curves of @xmath166 vs. @xmath156 for the different system sizes , @xmath132 , @xmath120 , @xmath121 and @xmath122 , all intersect at a common point , @xmath136 .
this is exactly the same value as found in our analysis of @xmath108 ( see fig.[f2 ] ) .
= 3.2truein we therefore conclude that @xmath166 gives a self - consistent determination of @xmath85 , and that this value is considerably larger than estimates one would get from consideration of @xmath177 .
in fact , estimates of a @xmath85 from @xmath177 all lie _ below _ the zero field critical temperature @xmath192 and _ decrease _ as @xmath193 _ increases _ , while the values determined from @xmath166 all lie _ above _ @xmath192 , and _ increase _ as @xmath193 _
increases_. if @xmath166 is indeed a scale invariant quantity , we can postulate that it should obey a scaling relation similar to @xmath103 , i.e. , @xmath194 based on our analysis of @xmath108 in section iii.b , we may expect @xmath7 . in fig.[f9 ]
we therefore show a scaling collapse of the data for @xmath180 from fig.[f7 ] , plotting @xmath166 vs. @xmath195l_{\perp}$ ] , where @xmath85 is determined by a best fit of the data to the scaling form . we find a reasonably good collapse for all sizes , for both aspect ratios , using a single value of @xmath196 .
= 3.2truein in fig.[f10 ] we show a similar scaling collapse of the data for @xmath116 from fig.[f8 ] . fitting the data for @xmath166 to a 4th order polynomial expansion of the scaling function
, we find an excellent collapse , for all system sizes , using the parameters @xmath197 and @xmath198 .
these results agree very well with the values obtained from the scaling analysis of @xmath108 , given in section iii.b .
the quality of the collapse is very much better here than it was for @xmath108 in fig.[f3 ] .
= 3.2truein finally , in analogy with the winding @xmath199 , we have also considered the probability @xmath200 to find a vortex path percolating through the system in the negative @xmath22 direction , opposite to the applied magnetic field .
we expect @xmath200 to obey a scaling relation similar to that of eq.([eqscaleo ] ) . to compute @xmath200 we have used tracing method ( iii ) in which we explicitly search through all possible connections to find any such paths .
we show our results for @xmath200 vs. @xmath131 for vortex density @xmath116 and anisotropy @xmath178 in figs.[f10.1 ] and [ f10.2 ] , for system aspect ratios @xmath179 and @xmath201 respectively . as with @xmath202 shown in figs.[f3.1 ] and [ f3.2 ] , the intersection points of the curves for different sizes appear to decrease in temperature as @xmath128 increases .
again , we can not say if this is a failure of our scaling hypothesis , or a failure to reach sufficiently large @xmath172 . also , analogous to our findings for the windings @xmath103 and @xmath203 , @xmath200 appears to be vanishing at a temperature _ above _ the @xmath204 where the curves of @xmath166 intersect .
= 3.2truein = 3.2truein
if @xmath85 , as determined by @xmath108 or @xmath166 , does indeed represent a true thermodynamic transition , we would expect to see some signature of this transition in more conventional thermodynamic quantities .
in the recent experiments of ref . , a step like anomaly in the specific heat @xmath205 was observed in the vortex line liquid region , reminiscent of an inverted mean field transition . in their
numerical simulations,@xcite nguyen and sudb claimed to see an anomalous glitch in the specific heat at the temperature they identified as @xmath85 from their calculation of @xmath177 .
however , this glitch corresponded to only a single data point very slightly displaced above an otherwise smooth background ; and in the previous section we have demonstrated that @xmath177 significantly underestimates @xmath85 , hence there is no reason to expect any anomaly in @xmath205 at that temperature .
in this section we report on high precision measurements of the specific heat @xmath205 , for the same parameters we have studied in the earlier sections . if the @xmath85 , as found using the vortex path tracing method ( ii ) , is indeed a true thermodynamic phase transition with critical exponent @xmath7 ( as our scaling analyses found ) , then hyperscaling would suggest a specific heat exponent of @xmath206 .
we thus do not expect to see a diverging @xmath205 , however some feature should be present . in fig.[f11 ]
we plot @xmath205 vs. @xmath156 , in the near vicinity of @xmath207 , for the same parameters @xmath116 , @xmath178 , and @xmath179 as used in figs.[f2 ] , [ f8 ] and [ f10 ] .
we show results for @xmath132 , @xmath120 , @xmath121 and @xmath122 , using from @xmath208 monte carlo passes through the lattice , depending on the system size .
we find no noticeable finite size dependence , and no hint of any feature at all , near the previously determined @xmath209 .
= 3.2truein in fig.[f12 ] we plot @xmath205 vs. @xmath156 , over a broad temperature range , for the same parameters @xmath180 and @xmath178 as used in figs.[f7 ] and [ f9 ] , but for a single large system size @xmath184 and @xmath210 .
again we see no hint of any anomaly near the previously determined @xmath196 .
we have carried out detailed monte carlo investigations of the 3d uniformly frustrated xy model in order to search for a proposed `` vortex loop blowout '' transition within the vortex line liquid phase of a pure extreme type - ii superconductor . such a transition had been predicted from general theoretical arguments by teanovi.@xcite evidence for such a transition was claimed in numerical simulations by sudb and co - workers,@xcite and in specific heat measurements on high purity ybco single crystals.@xcite we have made explicit measurements of the vortex line windings @xmath108 and @xmath203 , which are the key quantities in teanovi s theory .
we have re - examined sudb s calculation of the percolation probability @xmath211 .
our results raise several questions concerning teanovi s theory .
we have found that the values of @xmath108 and @xmath203 depend sensitively on the precise scheme one uses to trace out vortex line paths .
for the natural choice of random connectivity at vortex line intersections , both @xmath108 and @xmath203 appear to vanish at all temperatures as @xmath82 .
only when we specifically search first for percolating paths , when computing the windings , do we find that the windings converge to non zero values above a certain temperature . in this case , we find that the transverse winding @xmath108 obeys the finite size scaling form expected from teanovi s theory , however the critical exponent we find is @xmath7 , rather than the predicted @xmath212 of the inverted 3d xy transition . for the longitudinal winding @xmath203 we have been unable to find the expected scaling form . whether this is because @xmath203 does not scale , or because our systems are all too small to be in the scaling limit , we can not be certain .
it does appear that , upon cooling , @xmath203 vanishes at a temperature above that where @xmath108 vanishes .
this would be contrary to teanovi s theory .
however , since we have not succeeded to find scaling for @xmath203 , we can not be certain of knowing exactly where it vanishes as @xmath82 .
independent of teanovi s theory , it is natural to think that , as temperature and hence vorticity increases , the vortex lines may form percolating paths ( note however that the directedness of the vortex line segments , and the condition of divergenceless paths , means that this is no ordinary percolation problem ! ) .
we have therefore , following sudb and co - workers , searched explicitly for such percolating paths in the direction transverse to the applied magnetic field , as well as in the direction parallel but opposite to the applied magnetic field . defining transverse percolation as the existence of a vortex line path that extends entirely across the system in the direction transverse to the applied magnetic field _ without _ simultaneously extending entirely across the system in the parallel direction
, we have shown that sudb s procedure , which ignores the transverse periodic boundary conditions and does not require the percolating path to close upon itself , leads to inconsistent predictions for the transition temperature as one varies the system aspect ratio . only by requiring
that the transverse percolating path close upon itself , _ without _ ever winding in the parallel direction , do we find a consistent transition temperature independent of aspect ratio .
the percolation transition found this way agrees both in critical temperature @xmath85 and exponent @xmath144 with the results from our analysis of the transverse winding @xmath108 .
we have also computed the probability to find a percolating path in the direction parallel but opposite to the applied magnetic field . here , analogous to our results for @xmath203
, this negative @xmath22 percolation appears to occur at a temperature higher than that of the transverse percolation , however we have not succeeded to find a clear scaling of this parallel percolation probability .
note that the transverse percolation transition temperature @xmath213 that we find _ increases _ above the zero field transition temperature @xmath192 as the magnetic flux density @xmath193 _
increases_. this is in striking contrast to the conclusion of sudb and co - workers who proposed @xmath213 to _ decrease _ below @xmath192 as @xmath193 _
increases_. while our results do seem consistent with a well defined transverse percolation transition , one can ask if this is a purely geometrical feature of the vortex line paths , or whether it also corresponds to a true thermodynamic phase transition , i.e. something one could detect in a suitable thermodynamic derivative of the free energy . to investigate this question we have carried out high precision monte carlo measurements of the specific heat @xmath205 . our results for @xmath205 show no feature whatsoever near the percolation transition @xmath85 , nor
do we find any finite size effect .
in particular we see no evidence for a step like feature as was observed experimentally in ybco . to conclude , we have found evidence for a well defined transverse percolation temperature within the vortex line liquid phase of a model type - ii superconductor .
the connection between this transition and teanovi s theory of a vortex loop `` blowout '' transition remains unclear .
it also remains unclear whether or not this percolation transition has any observable thermodynamic manifestation .
we would like to thank prof .
z. teanovi and prof .
a. sudb for many helpful conversations .
this work was supported by the engineering research program of the office of basic energy sciences at the department of energy grant de - fg02 - 89er14017 , the swedish natural science research council contract no .
e 5106 - 1643/1999 , and by the resources of the swedish high performance computing center north ( hpc2n ) . travel between rochester and ume was supported by grants nsf int-9901379 and stint 99/976(00 ) .
x. hu , s. miyashita and m. tachiki , phys .
lett . * 79 * , 3498 ( 1997 ) and phys .
b * 58 * 3438 ( 1998 ) ; a. k. nguyen and a. sudb , phys .
b * 58 * , 2802 ( 1998 ) ; p. olsson and s. teitel , phys . rev
. lett . * 82 * , 2183 ( 1999 ) . for large systems , it is crucial to have an efficient algorithm to search for such paths .
we use the following .
first , all intersection points are located .
picking one such point at random , we trace out a path starting from this intersection point until it arrives at another intersection point .
if the height traveled from the starting to the new intersection point is @xmath214 , we stop this search and start again at a different intersection point ; if not , we continue tracing the path until then next intersection point is encountered and then repeat the height test . since most paths connect to field lines which travel in the @xmath215 direction , most such tracings are quickly aborted . however , since each possible transverse loop with @xmath50 contains an intersection point that is at the largest height of all intersection points on that loop , we are guarenteed to ultimately find this path with this search algorithm .
e. a. jagla and c. a. balseiro , phys .
b * 53 * , r538 ( 1996 ) ; _ ibid . _ * 53 * , 15305 . in these works
, the authors considered _ all _ transverse percolating loops , including those with net winding in the parallel direction , @xmath216 .
one can show that the onset of such loops , which in general involve the participation of the field induced lines and so do have @xmath216 , coincides with the vanishing of the longitudinal helicity modulus and so occurs at the melting @xmath217 , rather than the proposed @xmath85 . only by restricting to loops with @xmath218
does one probe @xmath85 . | the 3d uniformly frustrated xy model is simulated to search for a predicted `` vortex loop blowout '' transition within the vortex line liquid phase of a strongly type - ii superconductor in an applied magnetic field .
results are shown to strongly depend on the precise scheme used to trace out vortex line paths .
while we find evidence for a transverse vortex path percolation transition , no signal of this transition is found in the specific heat . |
the standard model ( sm ) @xcite provides a remarkably successful description of strong , weak and electromagnetic interactions at the energy scale up to @xmath9 .
it implies the conservation of baryon number @xmath10 and lepton number @xmath8 as the automatic consequence of gauge invariance and renormalizability . despite its tremendous success ,
the sm suffers from some conceptional difficulties , such as the hierarchy problem and the nonoccurrence of gauge coupling unification at high energy scale .
supersymmetry theories may provide solutions for such problems , and are widely expected as the most appealing extensions of the sm .
the simplest supersymmetric extension of the sm with minimal particle content is called minimal supersymmetric standard model ( mssm ) @xcite . to prevent proton from rapid decay @xcite ,
a multiplicative @xmath1-parity is introduced as @xmath11 in terms of @xmath10 , @xmath8 and spin quantum number @xmath12 @xcite , which leaves all the sm particles @xmath13 while for their superpartners @xmath14 .
the conservation of the @xmath1-parity imposes strict simultaneous @xmath10 and @xmath8 symmetry , and results in a stable lightest supersymmetric particle ( lsp ) , which can be a candidate for dark matter if it is color and electrically neutral .
however , such a stringent symmetry appears to be short of theoretical basis , especially as it is known that a stable proton can survive by requiring only either @xmath10 or @xmath8 conservation and partial lepton - flavor violation should be introduced to accommodate neutrino oscillations .
consequently , the @xmath1-parity - violating ( rpv ) interactions could be included into the most general representation of the superpotential as @xcite @xmath15 where @xmath16 are generation indices , while @xmath17 isospin and @xmath18 color indices are suppressed . @xmath19 and @xmath20 are the lepton and quark @xmath17 left - handed doublet chiral superfields , @xmath21 , @xmath22 and @xmath23 are the right - handed singlets , and @xmath24 is the left - handed up - type higgs doublet .
the bilinear terms @xmath25 , denoting the mixture of higgs and lepton superfields , will generate neutrino masses in a natural way .
the trilinear terms @xmath26 and @xmath3 with dimensionless rpv yukawa couplings @xmath27 and @xmath28 only violate @xmath8 symmetry , while the @xmath29-parameterized trilinear terms @xmath30 only break baryon - number conservation .
the constraints on the rpv couplings have been derived by precise measurements of low energy processes and decays , and are well summarized in the literature @xcite .
the most distinctive phenomenology of the @xmath1-parity - violating mssm ( @xmath31-mssm ) is that , the lsp is no longer stable and is not necessarily an electrically neutral and uncolored particle . as a consequence , supersymmetric scalar leptons and quarks , such as the lighter sbottom @xmath32 , can be candidates for the lsp , and the massive scalar fermion lsp can only decays into ordinary sm particles after its single / pair production at high erergy colliders . as we know , the bottom squark could be pair - produced at the large hadron collider ( lhc ) via qcd interactions .
the production rates and the differential distributions with respect to some kinematical variables , including the next - to - leading order qcd corrections @xcite and the electroweak contributions @xcite , have been calculated in several susy scenarios .
as shown in @xmath33 , the cross section for the sbottom pair production at the lhc decreases by @xmath34 orders of magnitude as the increment of @xmath35 from @xmath36 to about @xmath37 , due to the drastic reduction of the final state phase space and the smallness of parton densities in proton at large @xmath38 .
.[table0-locs - lhc ] leading order cross sections for sbottom - antisbottom production at the lhc .
the results for @xmath39 and @xmath40 are quoted from refs.@xcite and @xcite , respectively . [ cols="^,^,^ " , ] + as shown in @xmath41 , the relative corrections induced by these contributions to the rpv signal cross section and the luminosity required for the @xmath42 discovery of the lighter sbottom resonance , @xmath43 and @xmath44 , are almost less then @xmath45 .
therefore , the numerical results and conclusion of the paper are reasonable . *
the above lighter sbottom resonance searching strategy can be easily translated to probing the lighter scalar top @xmath46 ( assuming to be the lsp ) and the lepton - number - violating @xmath47-type yukawa interaction , via @xmath48 process in @xmath49 collision at the lhec . however , due to the lower luminosity of positron beam and the smaller fraction of energetic valence @xmath50-quark density in proton against electron beam and @xmath51-quark density , the potential of the discovery of the lighter stop resonance or the constraint on the relevant rpv coupling in @xmath49 scattering would be diminished , compared to the sbottom resonance search in @xmath6 collision mode .
in this paper , the possibility of probing the lepton - number - violating @xmath3 @xmath4-type interaction in the deep inelastic scattering at the lhec is investigated . under the single coupling dominance hypothesis that all the rpv couplings except @xmath4 are zero , the lighter sbottom can be produced and subsequently decay into electron and jet .
taking advantage of the enhancement of the narrow width resonant production of sbottom and distinctive kinematic distributions of @xmath52 decay products in the c.m .
frame of the two - body final state , an optimized event selection strategy can be developed to improve the sensitivity of the rpv signal over sm background . with
about @xmath5 integrated luminosity expected in @xmath2 @xmath6 collision at the lhec , either the lighter sbottom rpv resonance could be directly detected up to @xmath7 , or a new @xmath53 c.l .
upper bound on the @xmath8-violating coupling @xmath4 , which is one order of magnitude smaller than that obtained from the ckm matrix element @xmath54 , could be derived . | we investigate the resonant production of the lighter sbottom at the proposed large hadron electron collider and its subsequent decay into @xmath0 final state , under the single coupling dominance hypothesis in the framework of the @xmath1-parity - violating minimal supersymmetric standard model .
it is concluded that the lhec running at the high electron beam energy configuration of @xmath2 would provide excellent opportunities to search for the sbottom resonance and probe the lepton - number - violating @xmath3 @xmath4-type yukawa interaction . with about @xmath5 integrated luminosity expected in @xmath2 @xmath6 collision at the lhec ,
either the lighter sbottom rpv resonance could be directly detected up to @xmath7 , or the @xmath8-violating coupling @xmath4 could be constrained at an unprecedented level of precision compared with all the knowledge derived from indirect measurements .
= 0.32 in addtoresetequationsection |
the purpose of the paper is to expose potentially interesting relation of the weyl group associated with the color gauge @xmath0 symmetry to the structure of qcd vacuum . at the level of classical yang - mills
lagrangian the weyl group symmetry is trivial .
however , the vacuum structure of qcd is determined by quantum effects . the standard way to discuss the vacuum structure of
the theory in terms of effective quantum action relates to the landau - ginsburg type construction based on the symmetries of the theory .
we consider the landau - ginsburg lagrangian for pure yang - mills gauge fields invariant under the standard space - time and local gauge @xmath0 transformations .
it is demonstrated that a set of twelve degenerated minima of the action density exists as soon as a nonzero gluon condensate is postulated in the action .
the minima are connected to each other by the weyl group transformations associated with the color @xmath1 algebra and parity transformation .
the presence of degenerated discrete minima in the lagrangian leads to the solutions of the effective equations of motion in the form of the kink - like gauge field configurations interpolating between different minima . as an example
we write down the simplest solution which interpolates between self - dual and anti - self - dual abelian homogeneous fields and consider the spectrum of covariant derivative squared @xmath4 in the presence of this kink background field .
the kink configuration is seen here as a domain wall separating the regions with almost self - dual and anti - self - dual abelian gauge field .
it should be stressed that consideration itself and the results of this paper can be seen as instructive but very preliminary ones .
there are several observations and phenomenological estimates which can be considered as a qualitative motivation for introducing the effective action to be discussed in the next section .
the weyl group associated with @xmath5 gauge theory can be conveniently exposed in terms of representation of the gauge fields suggested in a series of papers by s. shabanov @xcite , y.m .
cho @xcite , l.d .
faddeev and a. j. niemi @xcite and , recently , by k .-
kondo @xcite . in this parameterization the abelian part @xmath6 of the gauge field @xmath7 is separated manifestly , @xmath8\hat{n } ( x)=b_\mu(x)\hat{n}(x ) , \nonumber \\ { \hat c}_\mu ( x ) & = & g^{-1}\partial_\mu \hat{n}(x)\times \hat{n}(x ) , \nonumber\\ { \hat x}_\mu ( x ) & = & g^{-1}{\hat n}(x ) \times \left ( \partial_\mu { \hat n}(x ) + g { \hat a}_\mu ( x ) \times { \hat n}(x ) \right ) , \nonumber\end{aligned}\ ] ] where @xmath9 , @xmath10 , @xmath11 , and @xmath12=if^{abc}t^c.\end{aligned}\ ] ] the field @xmath13 is seen as the abelian field in the sense that @xmath14=0 $ ] .
the comprehensive analysis of the rg - improved one - loop effective action for the abelian component @xmath15 with the constant @xmath16 and the covariantly constant ( anti)self - dual field @xmath17 was given long time ago by minkowski @xcite , pagels and tomboulis @xcite , and leutwyler @xcite .
the analysis based on the trace anomaly of the energy - momentum tensor and renormalization group leads to the following form of the effective potential for the @xmath18 gauge group @xcite : @xmath19.\end{aligned}\ ] ] here constant @xmath20 is defined as @xmath21 . for the strong field
@xmath22 this expression agrees with the result of the explicit one - loop calculation @xcite @xmath23.\end{aligned}\ ] ] equation ( [ leffrg ] ) indicates that the effective potential can have a minimum for nonzero strength @xmath20 only for the negative value of the parameter @xmath24 .
this parameter can be treated as the dielectric constant as @xmath25 .
one loop result ( [ leff1loop ] ) displays the strong field asymptotics @xmath26 of the effective lagrangian and , hence , its boundedness from below . in both calculations
the constant @xmath24 is a free parameter .
some knowledge about the sign of @xmath24 could be obtained from the lattice calculations .
a minimum of the effective lagrangian at nonzero field strength was reported in @xcite .
however the most interesting region of small field strength is the most difficult one for the lattice calculation , and , as the authors of @xcite stressed , this result should be taken into account with great care .
existence of nonzero gluon condensate can be considered as a general phenomenological argument in favor of nonzero value of the field strength at the minimum of the qcd effective action and , hence , the negative value of @xmath24 .
certainly , the ordered state corresponding to a plain constant field can not be considered as an appropriate approximation for qcd vacuum as it breaks all the symmetries of qcd at once . required disorder in the mean field
could be provided by an ensemble of gauge field configurations with the strength being constant almost everywhere but changing directions in space and color space as well as self- and anti - self - duality in small regions of space - time reminiscent of the domain walls .
phenomenological model of confinement , chiral symmetry breaking and hadronization based on the ensemble of abelian ( anti-)self - dual fields was developed in a series of papers @xcite .
the dominance of the domain structured gauge field configurations has been observed in the recent lattice calculations @xcite . in paper
@xcite an effective model of @xmath18 gauge theory for the domain wall formation was considered .
the center symmetry realization in lattice version of qcd is in the focus of these studies . in this paper
, we consider the effective lagrangian which displays another model for the domain wall formation based on cp and the weyl symmetry breakdown triggered by the trace anomaly or , equivalently , the nonzero gluon condensate .
consider the following effective lagrangian for the gauge fields satisfying the requirements of invariance under the gauge group @xmath0 and space - time transformations , @xmath27 where @xmath28 the gauge coupling constant is absorbed into the gauge field , @xmath29 .
the signs of the constants @xmath30 , @xmath31 and @xmath32 are chosen in such a way that the effective lagrangian is bounded from below and has a minimum at nonzero value of the field strength squared @xmath33 in terms of eq .
( [ leffrg ] ) the choice of sign of @xmath30 corresponds to the negative @xmath24 .
lagrangian ( [ leff ] ) contains the lowest order covariant derivatives and the effective potential which has a polynomial in @xmath34 form .
thus the field @xmath35 plays the role of the order parameter .
the presence of the term @xmath36 is of the crucial importance since the weyl group becomes manifest only in this and higher orders in field strength .
further increase in the polynomial order in ( [ leff ] ) does not change qualitatively the character of the weyl group realization .
the form of effective lagrangian ( [ leff ] ) is not the most general one .
our aim is to study an instructive example rather than to deal with the full problem in all its complexity .
namely , let us consider a set of fields @xmath37 with the abelian field strength of the following form @xmath38 where matrix @xmath39 is an element of cartan subalgebra in the adjoint representation @xmath40 for @xmath41 this field corresponds to the abelian part @xmath42 of the gauge field in the representation ( [ eqnsfaddeevniemkondo ] ) .
it is convenient to introduce the following notation : @xmath43 with this notation one arrives at the following formulae for traces : @xmath44 respectively , the effective potential takes the form @xmath45.\end{aligned}\ ] ] the potential ( [ ueff(b ) ] ) is invariant under transformations @xmath46 , @xmath47 , which can be seen as specific rotations of @xmath39 in cartan subalgebra .
these transformations lead to permutations of the eigenvalues of @xmath48 and , hence , do not change the traces of @xmath49 .
the permutations correspond to the weyl group associated with @xmath1 algebra , i.e. , the group of reflections of the roots of @xmath1 .
the effective potential is invariant with respect to parity transformation which results in the degeneracy of the self- and anti - self - dual fields corresponding to @xmath50 .
altogether there are twelve discrete global degenerated minima at the following values of the variables @xmath51 , @xmath52 and @xmath53 @xmath54 it should be stressed here that we have postulated in ( [ leff ] ) the minimum at nonzero value of the scalar gauge invariant field @xmath55 equivalent to the existence of nonzero gluon condensate , but the set of minima in @xmath52 and @xmath53 appeared as a consequence of the space - time and local gauge symmetries of the effective lagrangian ( [ leff ] ) .
inclusion of higher powers of @xmath56 does not change this picture qualitatively , but the presence of the term @xmath57 is crucial since dependence on @xmath53 appears starting the 6th order in @xmath56 .
the minimum of the effective potential in @xmath53 is achieved for the values @xmath58 corresponding to the boundaries of the weyl chambers in the root space of @xmath1 .
existence of the degenerate minima in the effective action related to the weyl group was reported earlier in @xcite and , in the context of the one - loop effective potential of @xmath5 gauge theory in @xcite .
a mechanism of gauge field localization on a domain wall within the framework of one - loop effective action for pure yang - mills theory was presented in @xcite .
it is well - known that the presence of the discrete global vacua in a system leads to the existence of kink - like solutions of the equations of motion .
these solutions describe field configurations interpolating between different vacua and can be treated as the domain walls between regions in @xmath59 with particular constant values of the parameters @xmath52 and @xmath53 from the set ( [ minima ] ) . in order to exemplify this statement ,
let us write down lagrangian in terms of the fields @xmath60 , @xmath61 and @xmath62 .
suppose @xmath63 .
then we have @xmath64 , \nonumber \\ & & \frac{1}{4 } \partial_\mu f^a_{\rho\sigma } \partial_\mu f^a_{\rho\sigma } = \frac{\lambda^2}{2}\left(\partial_\mu \mathbf { h } \partial_\mu \mathbf{h } + \partial_\mu \mathbf{e } \partial_\mu\mathbf{e } + b_{\rm vac}^2 \ \partial_\mu \xi\ \partial_\mu \xi \right).\end{aligned}\ ] ] here @xmath65 is the minimal value of the effective potential corresponding to the constant values of @xmath51 , @xmath52 and @xmath53 given in ( [ minima ] ) . to self - dual at @xmath66 configuration : @xmath67 , @xmath68 , @xmath69
. here @xmath70 , @xmath71.,width=264 ] in order to separate the relevant variable @xmath52 from other degrees of freedom of * e * and * h * it is convenient to represent the electric field as @xmath72 where @xmath73 is a local rotation about unit vector * m * orthogonal to * h * , @xmath74 here @xmath75 is azimuthal angle of * h * with respect to @xmath76 , and * v * is a constant unit vector .
we will take @xmath77 after some algebraic transformations the lagrangian density @xmath78 takes the form @xmath79 let @xmath80 , @xmath81 and @xmath82 , then @xmath83 and the euler - lagrange equation @xmath84 let us look for solutions @xmath52 which depend only on one of the coordinates , say @xmath85 .
equation ( [ singordonextended ] ) takes the form of sin - gordon equation @xmath86 with kink solution @xmath87 according to eq.([kink ] ) , the angle between chromoelectric and chromomagnetic fields @xmath88 varies from @xmath89 to @xmath90 for @xmath91 $ ] .
it corresponds to the change from anti - self - dual to self - dual gauge field configuration , as is shown in fig.[comp ] similarly , if @xmath92 , @xmath81 and @xmath82 , then @xmath93 with the solution @xmath94+\frac{\pi k}{3 } , \ k=1,\dots,6,\end{aligned}\ ] ] interpolating between two consequent vacuum values of @xmath53 in ( [ minima ] ) associated with the boundaries of the @xmath95-th weyl chamber .
in this section , we estimate the change in the spectrum of color charged scalar field caused by the kink - like defect in @xmath52 in comparison with the spectrum in the presence of confining ( anti-)self - dual purely homogeneous abelian background . here
we consider the infinitely thin domain wall for @xmath52 which corresponds to @xmath96 in eq .
( [ kink ] ) .
since the kink interpolates between the cp conjugated vacua and some particular vacuum value of angle @xmath53 it is sufficient to consider the eigenvalue problem @xmath97 in the case of infinitely thin wall , the field @xmath98 is self - dual for @xmath99 , anti - self - dual for @xmath100 , but inside the domain wall ( at @xmath101 ) electric and magnetic fields are orthogonal to each other .
inside the domain bulk vector potential can be represented as the homogeneous self- or anti - self - dual field @xmath102 squire integrable solution is well known in this case .
the following field strength configuration can be chosen without loss of generality @xmath103 equation ( [ eigeneqn ] ) is equivalent to @xmath104 \phi = \frac{\lambda}{4b } \phi,\end{aligned}\ ] ] where creation and annihilation operators @xmath105 , @xmath106 , @xmath107 , @xmath108 are expressed in terms of the operators @xmath109 , @xmath110 : @xmath111 here `` @xmath112 '' indicates the self - dual and anti - self - dual configuration . the eigenvalues and the square integrable eigenfunctions are @xmath113 where @xmath114 for self - dual field , @xmath115 for anti - self - dual field .
the spectrum is discrete . at the domain wall
the eigen functions are continuous .
there is an infinite degeneracy of the eigen values . inside the domain wall ( @xmath101 )
vector potential can be chosen as @xmath116 charged field displays continuous spectrum similar to landau levels .
square integrable over @xmath117 eigen functions take the form @xmath118 where functions @xmath119 @xmath120 are solutions of the eigen value problem @xmath121\chi_n=\lambda_n \chi_n,\\\end{aligned}\ ] ] with the eigen values @xmath122 the character of charged field modes is qualitatively different in the domain bulk ( self - dual field ) and inside the domain wall ( crossed electric and magnetic fields ) , which illustrates the character of the problem to be solved to obtain a continous common solution for the domain bulk and wall for the case of the finite width of the kink .
the form of the eigen functions ( [ efosc ] ) indicates charge field confinement in the bulk and the presence of `` plain wave '' solutions inside the wall .
in terms of the effective lagrangian we investigated manifestations of cp and the weyl group associated with the @xmath0 gauge theory . it is shown that the requirement of nonzero gluon condensate leads to the existence of a set of degenerated minima and , as a consequence , triggers the kink - like gauge field configurations interpolating between different minima .
the spectrum of a charged scalar field in the background of the kink - like fields was estimated .
the bound state form of the eigen functions ( [ efosc ] ) indicates confinement of charged field inside domain , while the `` plain wave '' eigen modes ( [ lpv ] ) exist inside the wall .
the eigenfunction properties and the propagator of a charged field in the kink background have to be studied in detail for the case of finite width of the kink .
it is important to investigate the eigenvalue problem for fermionic charged fields and the chiral symmetry realization in the kink - like background . the domain model of qcd vacuum developed in @xcite
is based on the ensemble of the background gluon fields with the field strength being constant almost everywhere in @xmath59 .
the direction of the field in space and color space as well as duality of the field are random parameters of the domains .
all configurations of this type are summed up in the partition function .
the domain model exhibits confinement of static and dynamic quarks , spontaneous breaking of the flavor chiral symmetry , @xmath123 symmetry is broken due to the axial anomaly , strong @xmath124 violation is absent in the model .
the domain boundaries were introduced by means of bag - like boundary conditions imposed on the gluon and quark fluctuation fields , which made the model unbalanced and considerably complicated all calculations .
gauge field configurations investigated in the present paper provides us with an interesting option for parameterization of the domain structured ensemble of gluon fields .
+ * acknowledgments : * we would like to thank g. efimov , a. isaev , a. dorokhov , a. khvedelidze and n. kochelev for discussions and valuable comments . _
cho _ // phys .
44 ( 1980 ) 1115 .
_ y. m. cho , j. h. kim and d. g. pak_// mod .
lett . a 21 ( 2006 ) 2789 .
shabanov _ // j. math .
43 ( 2002 ) 4127 [ hep - th/0202146 ] .
_ s. v. shabanov _ //
326 ( 2000 ) 1 [ arxiv : hep - th/0002043 ] ; _ s. v. shabanov and j. r. klauder _ // phys
b 456 ( 1999 ) 38 . _
l. v. prokhorov _ //
fiz . , 35 ( 1982 ) 229 .
_ l. d. faddeev , a. j. niemi _ //
2007 ; _ ibid _ , phys .
b 449 ( 1999 ) 214 . _
kri - ichi kondo , toru shinohara , takeharu murakami _ //
. phys . 120
( 2008 ) 1 [ arxiv:0803.0176 [ hep - th ] ] .
_ p. minkowski _ //
b 76 ( 1978 ) 439 ; _ ibid _ nucl . phys .
b177 ( 1981 ) 203 .
_ h. pagels , and e. tomboulis _ //
b 143 ( 1978 ) 485 . _
h. leutwyler _
b96 ( 1980 ) 154 ; _ ibid _ // nucl . phys .
b 179 ( 1981 ) 129 . _
h. d. trottier and r. m. woloshyn_// phys .
70 ( 1993 ) 2053 .
efimov , and s.n .
nedelko_// phys .
d51 ( 1995 ) 176 ; _ ya .
v. burdanov , et .
al_// phys .
d 54 ( 1996 ) 4483 .
kalloniatis and s.n .
nedelko _ //
d 64 ( 2001 ) 114025 ; _ a.c . kalloniatis and s.n .
nedelko _ //
d 64 ( 2001 ) 114025 ; _ ibid_// phys . rev .
d 66 ( 2002 ) 074020 ; _ ibid _ // phys .
d 69 ( 2004 ) 074029 ; _ erratum _ phys .
d 70 ( 2004 ) 119903 ; _ ibid _ // phys
d 71 ( 2005 ) 054002 ; _ ibid _ // phys .
d 73 ( 2006 ) 034006 . _
p. j. moran and d. b. leinweber _ //
arxiv:0805.4246 [ hep - lat ] .
_ p. j. moran and d. b. leinweber_// pos * lat2007 * ( 2007 ) 383 [ arxiv:0710.2380 [ hep - lat ] ] .
ph . de forcrand , a. kurkela and a. vuorinen_// phys .
d 77 ( 2008 ) 125014 . _
de forcrand_// aip conf .
( 2007 ) 29 [ arxiv : hep - lat/0611034 ] .
_ e. m. ilgenfritz , et al_// pos * lat2007 * ( 2007 ) 311 [ arxiv:0710.2607 [ hep - lat ] ] .
a. kobakhidze , arxiv:0807.4578 [ hep - th ] ( 2008 ) . | effective lagrangian for yang - mills gauge fields invariant under the standard space - time and local gauge @xmath0 transformations is considered .
it is demonstrated that a set of twelve degenerated minima exists as soon as a nonzero gluon condensate is postulated .
the minima are connected to each other by the parity transformations and weyl group transformations associated with the color @xmath1 algebra .
the presence of degenerated discrete minima in the effective potential leads to the solutions of the effective euclidean equations of motion in the form of the kink - like gauge field configurations interpolating between different minima .
spectrum of charged scalar field in the kink background is discussed .
+ pacs numbers : 12.38.aw , 12.38.lg , 14.70.dj * weyl group , cp and the kink - like field configurations in the effective @xmath0 gauge theory * b.v .
galilo @xmath2 , s.n .
nedelko @xmath3 + + |
the study of the properties of baryon resonances is entering a new era with the forthcoming new and more accurate data from new facilities , such as jefferson lab .
, mami , elsa and brookhaven .
effective models of baryons which are based on three constituents ( @xmath0 ) share a common spin - flavor - color structure but differ in their assumptions on the spatial dynamics . for example , quark potential models in nonrelativistic @xcite or relativized @xcite forms emphasize the single - particle aspects of quark dynamics for which only a few low - lying configurations in the confining potential contribute significantly to the eigenstates of the hamiltonian . on the other hand , some regularities in the observed spectra , such as linear regge trajectories and parity doubling , hint that an alternative , collective type of dynamics may play a role in the structure of baryons . in this contribution
we discuss a collective model within the context of an algebraic approach @xcite and present some results for electromagnetic form factors @xcite and strong decay widths @xcite .
1.75 in we consider a collective model in which the baryon resonances are interpreted in terms of rotations and vibrations of the string configuration in fig .
[ rb : geometry ] .
a fit to the 3 and 4 star nucleon and delta resonances gives a r.m.s .
deviation of 39 mev @xcite .
the corresponding oblate top wave functions are spread over many oscillator shells and hence are truely collective .
the baryon wave functions have the form @xmath1_{(v_1,v_2);k } \ , \right > ~. \label{wf}\end{aligned}\ ] ] the spin - flavor part has the usual @xmath2 classification and determines the permutation symmetry of the state .
the spatial part is characterized by the labels : @xmath3 , where @xmath4 denotes the vibrations ( stretching and bending ) of the string configuration in fig . [ rb : geometry ] ; @xmath5 denotes the projection of the rotational angular momentum @xmath6 on the body - fixed symmetry - axis and @xmath7 the parity .
finally , @xmath8 and @xmath9 are the spin and the total angular momentum @xmath10 . in this contribution
we focus on the nucleon resonances which are interpreted as rotational excitations of the @xmath11 vibrational ground state .
the electromagnetic ( strong ) coupling is assumed to involve the absorption or emission of a photon ( elementary meson ) from a single constituent . the collective form factors and decay widths
are obtained by folding with a probability distribution for the charge and magnetization along the strings of fig .
[ rb : geometry ] @xmath12 where @xmath13 is a radial coordinate and @xmath14 is a scale parameter . in the algebraic approach
these form factors and decay widths can be obtained in closed analytic form ( in the limit of a large model space ) which allows us to do a straightforward and systematic analysis of the experimental data .
the ansatz of eq .
( [ gbeta ] ) for the probability distribution is made to obtain a dipole form for the proton electric form factor @xmath15 the same distribution is used to calculate transition form factors and decay widths . as a result , all collective form factors are found to drop as powers of the momentum transfer @xcite .
this property is well - known experimentally and is in contrast with harmonic oscillator based quark models in which all form factors fall off exponentially .
electromagnetic inelastic form factors can be measured in electroproduction of baryon resonances . they are expressed in terms of helicity amplitudes @xmath16 , where @xmath17 indicates the helicity and @xmath18 represents proton ( @xmath19 ) or neutron ( @xmath20 ) couplings . in a string - like model of hadrons
one expects @xcite on the basis of qcd that strings will elongate ( hadrons swell ) as their energy increases .
this effect can be easily included in the present analysis by making the scale parameters of the strings energy- dependent .
we use here the simple ansatz @xmath21 where @xmath22 is the nucleon mass and @xmath23 the resonance mass .
this ansatz introduces a new parameter ( @xmath24 ) , the stretchability of the string .
5.5 cm in fig .
[ rb : d13 ] we show the transverse helicity amplitudes @xmath25 for the @xmath26 resonance calculated in the breit frame ( a factor of @xmath27 is suppressed ) .
the scale parameter @xmath14 is determined in a simultaneous fit to the nucleon charge radii and the nucleon elastic form factors : @xmath28 fm @xcite .
the effect of stretching on the helicity amplitudes is sizeable ( especially if one takes the value @xmath29 which is suggested by qcd arguments @xcite and the regge behavior of nucleon resonances ) .
the data for @xmath26 ( and also for @xmath30 ) show a clear indication that the form factors are dropping faster than expected on the basis of the dipole form .
the disagreement at low momentum transfer may be due to the neglect of the meson cloud .
[ rb : strong ] .@xmath31 and @xmath32 decay widths of 3 and 4 star nucleon resonances in mev . [ cols="<,^,^,^,^,^,^ " , ] in addition to electromagnetic couplings , strong decays provide an important , complementary , tool to study the structure of baryons .
we consider decays with emission of @xmath33 and @xmath34 .
the experimental data @xcite are shown in table [ rb : strong ] , where they are compared with the results of our calculation .
the calculated values depend on two parameters @xmath35 and @xmath36 in the transition operator and on the scale parameter @xmath14 of eq .
( [ gbeta ] ) .
these parameters are determined from a least - square fit to the @xmath31 partial widths ( which are relatively well known ) with the exclusion of the @xmath37 resonances whose assignments are not clear due to possible mixing between the @xmath38 and @xmath39 and/or the possible presence of a third @xmath37 resonance @xcite .
the calculation of @xmath31 decay widths is found to be in fair agreement with experiment ( see table [ rb : strong ] ) .
these results are to a large extent a consequence of spin - flavor symmetry .
the calculated widths for the @xmath32 channel are systematically small .
we emphasize here that , since the transition operator was determined from the @xmath31 decays , the @xmath34 decays are calculated without introducing any further parameters .
the results of this analysis suggest that the large @xmath34 width for the @xmath40 is not due to a conventional @xmath41 state .
one possible explanation is the presence of another state in the same mass region , _
e.g. _ a quasi - bound meson - baryon @xmath8 wave resonance just below or above threshold , for example @xmath42 , @xmath43 or @xmath44 @xcite .
another possibility is an exotic configuration of four quarks and one antiquark ( @xmath45 ) .
we have analyzed simultaneously electromagnetic form factors and strong decay widths in a collective model of the nucleon .
the helicity amplitudes are folded with a probability distribution for the charge and magnetization , which is determined from the well - established dipole form of the nucleon electromagnetic form factors .
the same distribution function is used to calculate the transition form factors for the excited baryons . as a result ,
all form factors drop as powers of the momentum transfer . for electromagnetic couplings
we find that the inclusion of the stretching of baryons improves the calculation of the helicity amplitudes for large values of the momentum transfer .
the disagreement for small values of the momentum transfer @xmath46 ( gev / c)@xmath47 may be due to coupling of the photon to the meson cloud ,
_ i.e. _ configurations of the type @xmath48 . since such configurations have much larger spatial extent than @xmath41
, their effects are expected to drop faster with momentum transfer than the constituent form factors .
an analysis of the strong decay widths into the @xmath31 and @xmath32 channels shows that the @xmath33 decays follow the expected pattern .
our calculations do not show any indication for a large @xmath34 width , as is observed for the @xmath40 resonance .
the observed large @xmath34 width indicates the presence of another configuration , which is outside the present model space .
the results reported in this article are based on work done in collaboration with f. iachello ( yale ) .
the work is supported in part by grant no .
94 - 00059 from the united states - israel binational science foundation ( bsf ) , jerusalem , israel ( a.l . ) and by conacyt ,
mxico under project 400340 - 5 - 3401e and dgapa - unam under project in105194 ( r.b . ) . | we discuss recent calculations of electromagnetic form factors and strong decay widths of nucleon and delta resonances .
the calculations are done in a collective constituent model of the nucleon , in which the baryons are interpreted as rotations and vibrations of an oblate top . |
the work was supported in part by the russian foundation for basic research under the grant no .
04 - 02 - 16253 , and by the council on grants by the president of russian federation for the support of young russian scientists and leading scientific schools of russian federation under the grant no .
nsh-6376.2006.2 .
99 q. r. ahmad _ et al . _
( sno collab . ) , _ phys .
lett . _ * 87 * , 071301 ( 2001 ) ; _ ibid_. * 89 * , 011301 ; 011302 ( 2002 )
. y. fukuda _
( super - kamiokande collab . ) , _ phys .
* b433 * , 9 ( 1998 ) ; _ ibid_. * b436 * , 33 ( 1998 ) ; _ phys . rev
. lett . _ * 82 * , 2644 ( 1999 ) .
k. eguchi _
( kamland collab . ) , _ phys .
* 90 * , 021802 ( 2003 ) .
b. pontecorvo , _ zh . eksp .
_ * 33 * , 549 ( 1957 ) [ _ sov
. jetp _ * 6 * , 429 ( 1957 ) ] ; _ zh
fiz . _ * 34 * , 247 ( 1958 ) [ _ sov .
. jetp _ * 7 * , 172 ( 1958 ) ]
. l. wolfenstein , _ phys .
rev . _ * d17 * , 2369 ( 1978 ) .
d. ntzold and g. raffelt , _ nucl .
* b307 * , 924 ( 1988 ) .
p. b. pal and t. n. pham , _ phys .
* d40 * , 259 ( 1989 ) .
j. f. nieves , _ phys .
rev . _ * d40 * , 866 ( 1989 ) .
a. lobanov and a. studenikin , _ phys .
lett . _ * b564 * , 27 ( 2003 ) .
a. lobanov and a. studenikin , _ phys .
lett . _ * b601 * , 171 ( 2004 ) .
a. studenikin and a. ternov , _ phys .
lett . _ * b608 * , 107 ( 2005 )
. a. e. lobanov , _ phys .
lett . _ * b619 * , 136 ( 2005 ) .
a. grigoriev , a. studenikin and a. ternov , _ phys .
lett . _ * b622 * , 199 ( 2005 ) .
w. grimus and h. neufeld , _ phys .
lett . _ * b315 * , 129 ( 1993 ) .
dolivo , j.f .
nieves and p.b .
pal , _ phys .
* b365 * , 178 ( 1996 ) .
a.n . ioannisian and g.g .
raffelt , _ phys .
rev . _ * d55 * , 7038 ( 1997 ) .
e. braaten and d. segel , _ phys .
* d48 * , 1478 ( 1993 ) .
p. elmfors , d. grasso and g.raffelt , _ nucl .
phys . _ * b479 * , 3 ( 1996 ) . | the conversion @xmath0 of a neutrino with a magnetic moment is considered , caused by the additional wolfenstein energy acquired by a left - handed neutrino in medium , with an accurate taking account of the photon @xmath1 dispersion in medium .
it is shown that the threshold arises in the process , caused by the photon ( plasmon ) effective mass .
this threshold leaves no room for the so - called `` neutrino spin light '' in the most of astrophysical situations .
+ pacs nos . : 13.15.+g , 95.30.cq _ submitted to modern physics letters a _
the most important event in neutrino physics of the last decades was the solving of the solar neutrino problem , made in the unique experiment on the heavy - water detector at the sudbury neutrino observatory @xcite .
this experiment , together with the atmospheric and the reactor neutrino experiments @xcite , has confirmed the key idea by b. pontecorvo on neutrino oscillations @xcite . the existence of non - zero neutrino mass and lepton mixing is thereby established .
the sun appeared in this case as a natural laboratory for investigations of neutrino properties .
there exists a number of natural laboratories , the supernova explosions , where gigantic neutrino fluxes define in fact the process energetics .
it means that microscopic neutrino characteristics , such as the neutrino magnetic moment , the neutrino dispersion in an active medium , etc . , would have a critical impact on macroscopic properties of these astrophysical events .
this is the reason for a growing interest to neutrino physics in an external active medium . in an astrophysical environment
, the main medium influence on neutrino properties is defined by the additional wolfenstein energy @xmath2 acquired by a left - handed neutrino @xcite .
the general expression for this additional energy of a left - handed neutrino with the flavor @xmath3 is @xcite @xmath4 , \label{eq : enulgen}\end{aligned}\ ] ] where the functions @xmath5 are the number densities of background electrons , protons , neutrons , and neutrinos , and @xmath6 are the densities of the corresponding antiparticles . to find the additional energy for antineutrinos , one should change the total sign in the right - hand side of eq .
( [ eq : enulgen ] ) . for a typical astrophysical medium , except for the early universe and a supernova core
, one has @xmath7 , and @xmath8 , where @xmath9 is the barion density .
one obtains @xmath10 as @xmath11 , the additional energy acquired by muon and tau left - handed neutrinos is always negative . at the same time , the additional energy of electron left - handed neutrinos becomes positive at @xmath12 . and
vice versa , the additional energy for electron antineutrinos is positive at @xmath13 , while it is always positive for the muon and tauon antineutrinos .
on the other hand , right - handed neutrinos and their antiparticles , left - handed antineutrinos , being sterile with respect to weak interactions , do not acquire an additional energy .
the additional energy @xmath2 from eq .
( [ eq : enule ] ) gives an effective mass squared @xmath14 to the left - handed neutrino , @xmath15 where @xmath16 is the neutrino four - momentum in medium , while @xmath17 would form the neutrino four - momentum in vacuum , @xmath18 .
given a @xmath19 interaction , the additional energy of left - handed neutrinos in medium opens new kinematical possibilities for the radiative neutrino transition : @xmath20 it should be self - evident , that the influence of the substance on the photon dispersion must be taken into account , @xmath21 , where @xmath22 is the refractive index .
first , a possibility exists that the medium provides the condition @xmath23 ( the effective photon mass squared is negative , @xmath24 ) which corresponds to the well - known effect @xcite of `` neutrino cherenkov radiation '' .
in this situation , the neutrino dispersion change under the medium influence is being usually neglected , because the neutrino dispersion is defined by the weak interaction while the photon dispersion is defined by the electromagnetic interaction .
pure theoretically , one more possibility could be considered when the photon dispersion was absent , and the process of the radiative neutrino transition @xmath25 would be caused by the neutrino dispersion only . as the left - handed neutrino dispersion is only changed , transitions become possible caused by the @xmath19 interaction with the neutrino chirality change , e.g. due to the neutrino magnetic dipole moment .
just this situation called the `` spin light of neutrino '' ( @xmath26 ) , was first proposed and investigated in detail in an extended series of papers @xcite . however , in the analysis of this effect the authors overlooked such an important phenomenon as plasma influence on the photon dispersion . as will be shown below
, this phenomenon closes the @xmath26 effect for all real astrophysical situations . in this letter , we reanalyse the process @xmath27 taking into account both the neutrino dispersion and the photon dispersion in medium . having in mind possible astrophysical applications , it is worthwhile to consider the astrophysical plasma as a medium , which transforms the photon into the plasmon , see e.g. ref .
@xcite and the papers cited therein . to perform a kinematical analysis ,
it is necessary to evaluate the scales of the values of the left - handed neutrino additional energy @xmath2 and of the photon ( plasmon ) effective mass squared @xmath28 .
one readily obtains from eq .
( [ eq : enule ] ) : @xmath29 where the scale of the barion number density is taken , which is typical e.g. for the interior of a neutron star . on the other hand , a plasmon acquires in medium an effective mass @xmath30 which is approximately constant at high energies . for the transversal plasmon ,
the value @xmath28 is always positive , and is defined by the so - called plasmon frequency . in the non - relativistic classical plasma ( i.e. for the solar interior )
one has : @xmath31 for the ultra - relativistic dense matter one has : @xmath32 in the case of hot plasma , when its temperature is the largest physical parameter , the plasmon mass is : @xmath33 one more physical parameter , a great attention was payed to in the @xmath26 analysis @xcite , was the neutrino vacuum mass @xmath34 . as the scale of neutrino vacuum mass could not exceed essentially a few electron - volts , which is much less than typical plasmon mass scales for real astrophysical situations , see eqs .
( [ eq : omega_pl_nr])-([eq : omega_pl_h ] ) , it is reasonable to neglect @xmath34 in our analysis .
thus , in accordance with ( [ eq : ml ] ) , a simple condition for the kinematic opening of the process @xmath27 is : @xmath35 this means that the process becomes kinematically opened when the neutrino energy exceeds the threshold value , @xmath36 let us evaluate these threshold neutrino energies for different astrophysical situations . for the solar interior @xmath37 , @xmath38 , and the threshold neutrino energy is @xmath39 to be compared with the upper bound @xmath40 20 mev for the solar neutrino energies . for the interior of a neutron star , where @xmath41 , the wolfenstein energy for neutrinos ( [ eq : enule ] ) , ( [ eq : enulmu ] ) is negative , and the process
@xmath27 is closed . on the other hand
, there exists a possibility for opening the antineutrino decay . taking for the estimation @xmath42 , one obtains from ( [ eq : w ] ) and ( [ eq : omega_pl_r ] ) the threshold value @xmath43 to be compared with the typical energy @xmath40 mev of neutrinos emitted via the urca processes . for the conditions of a supernova core
, the additional energy of left - handed electron neutrinos can be obtained from eq .
( [ eq : enulgen ] ) as follows : @xmath44 where @xmath45 describes the fraction of trapped electron neutrinos in the core , @xmath46 .
taking typical parameters of a supernova core , we obtain @xmath47 to be compared with the averaged energy @xmath48 mev of trapped neutrinos . in the early universe , when plasma was almost charge symmetric , the wolfenstein formula ( [ eq : enulgen ] ) giving zero
should be changed to a more accurate expression for the additional energy which is identical for both neutrinos and antineutrinos @xcite @xmath49 the minus sign unambiguously shows that in the early universe , in contrast to the neutron star interior , the decay process is forbidden both for neutrinos and antineutrinos .
thus , the above analysis shows that the nice effect of the `` neutrino spin light '' , unfortunately , has no place in real astrophysical situations because of the photon dispersion . the sole possibility for the discussed process @xmath27 to have any significance could be connected only with the situation when an ultra - high energy neutrino threads a star .
obviously it could have only a methodical meaning .
let us calculate the process width for these purposes correctly .
a neutrino having a magnetic moment @xmath50 interacts with photons , and the lagrangian of this interaction is @xmath51 where @xmath52 , and @xmath53 is the tensor of the photon electromagnetic field . with the lagrangian ( [ eq : l ] ) ,
the invariant amplitude squared for the process @xmath27 , summarized over the transversal plasmon polarizations , can be obtained by the standard way : @xmath54 , \label{eq : m^2}\ ] ] where @xmath55 is the plasmon energy , @xmath56 is the angle between the initial neutrino momentum @xmath57 and the plasmon momentum @xmath58 .
it should be stressed that discussing ultra - high energy neutrinos , and consequently the high plasmon energies , one can consider with a good accuracy the plasmon mass @xmath30 as a constant depending on the plasma properties only , see eqs .
( [ eq : omega_pl_nr])-([eq : omega_pl_h ] ) .
this is in contrast to the left - handed neutrino effective mass squared @xmath14 , which is the dynamical parameter , see eq .
( [ eq : ml ] ) . the differential width of the process @xmath27 is defined as : @xmath59 where the plasmon energy @xmath55 can not be taken the vacuum one ( @xmath60 ) , as it was done in the @xmath26 analysis @xcite , but it is defined by the dispersion in plasma , @xmath61 .
performing a partial integration in eq .
( [ eq : dgamma ] ) , one obtains for the photon ( i.e. transversal plasmon ) spectrum @xmath62 where @xmath63 is the bohr magneton , and the notations are used @xmath64 , and @xmath65 . recall that @xmath66 is the threshold neutrino energy for the process to be opened . in fig .
[ fig : spectrum ] the function @xmath67 is shown for some values of the ratio @xmath68 . defining the spectrum of plasmons from the left - handed neutrino decay for different values of the ratio @xmath65 of the neutrino energy to the threshold neutrino energy , @xmath69 ( solid line ) , @xmath70 ( dotted line ) , and @xmath71 ( dashed line).,scaledwidth=60.0% ] instead of the photon energy spectrum ( [ eq : dgamma2 ] )
, one can obtain also the spatial distribution of final photons .
as the analysis shows , all the photons are created inside the narrow cone with the opening angle @xmath72 , @xmath73 performing the final integration in eq .
( [ eq : dgamma2 ] ) , one obtains the total width of the process @xmath74 .
\label{eq : gamma}\end{aligned}\ ] ] it should be noted that in the situation when @xmath75 , and the transition @xmath27 is forbidden , the crossed channel @xmath76 becomes kinemalically opened .
as the analysis shows , the plasmon spectrum and the total decay width are described in this case by the same eqs .
( [ eq : dgamma2 ] ) and ( [ eq : gamma ] ) , with the only substitution @xmath77 . to illustrate the extreme weakness of the effect considered ,
let us evaluate numerically the mean free path of an ultra - high energy neutrino with respect to the radiative decay , when the neutrino threads a neutron star .
for the typical neutron star parameters , @xmath78 , @xmath79 , we obtain @xmath80^{-1 } \ , , \label{eq : path}\ ] ] where the neutrino energy @xmath81 , @xmath82 tev is the threshold energy for such conditions .
the mean free path ( [ eq : path ] ) should be compared with the neutron star radius @xmath83 cm . in conclusion , we have shown that the effect of the `` neutrino spin light '' has no place in real astrophysical situations because of the photon dispersion in plasma .
the photon ( plasmon ) effective mass causes the threshold which leaves no room for the process . for a pure theoretical situation
when an ultra - high energy neutrino threads a star , the total width of the process @xmath27 is calculated with a correct taking account of the photon dispersion in plasma .
the extreme weakness of the effect considered is illustrated . |
the extragalactic gamma - ray sky at energies above 100 mev is mainly populated by blazars , radio loud active galactic nuclei ( agns ) whose emission is dominated by the relativistically boosted non - thermal continuum produced in a relativistic jet ( bulk lorentz factor @xmath0 ) pointing ( typical angles less than 5 degrees ) toward the earth .
blazars are among the most powerful and violently variable sources in the universe : their emission covers all the electromagnetic spectrum , from the radio to the gamma - ray band , and their bolometric output ( often concentrated in the gamma - ray band ) can exceeds @xmath1 erg s@xmath2 . the variability timescale can be as short as few minutes@xcite@xcite .
the interest for these sources is driven by the possibility to probe the innermost regions and the surroundings of the relativistic jets , close to the acceleration and collimation region ( @xmath3 gravitational radii ) . in the past decade
egret onboard _
cgro _ detected almost 70 blazars .
of the 106 sources in the recently released list@xcite ( lbas ) of the agns detected with high significance by the large area telescope ( lat ) onboard the _ fermi gamma ray telescope_@xcite in the first three months of operation ( august - october 2008 ) , 104 are blazars , the remaining two sources being the radiogalaxies perseus a and centaurus a. note that one of the sources classified as a blazar is a radio loud narrow line seyfert 1 galaxy , the first object of this class revealed at gamma rays@xcite@xcite .
the gamma - ray emission can extend up to the vhe band ( @xmath4 gev ) , at which the observations are performed by the cherenkov telescopes on the ground .
also in this band blazars dominate the extragalactic sky .
the vhe emission can be used to derive limits on the density of the ir - optical backgrounds exploiting the modification of the spectrum due to the absorption through the pair production process@xcite@xcite . due to the limited space , in the following
i will focus the attention on some specific issues : i ) the blazar sequence and its physical meaning ; ii ) the high - energy view of the the bl lac objects and some problems possibly requiring a revision of the simple one - zone emission models .
the spectral energy distribution ( sed ) of blazars displays two broad peaks , interpreted as due to synchrotron and inverse compton ( ic ) mechanisms in the popular leptonic models@xcite@xcite ( see , e.g. , ref . for alternatives hadronic models ) . as shown almost ten years ago by fossati et al.@xcite ,
averaged seds of blazars in different bins of radio luminosity show a trend with the radio ( or the bolometric ) luminosity : specifically , flat spectrum radio quasars ( fsrqs ) , the most powerful blazars , showing intense broad emission lines and sometimes a thermal bump in their optical spectra , have both peaks located at low frequencies , submm - ir and mev for the synchrotron and ic peak , respectively . on the contrary , in the less powerful bl lac objects ( characterized by very weak emission lines ) , the two bumps peak at uv - x - ray frequencies and above 100 gev . the existence and
the validity of such a trend ( dubbed `` blazar sequence '' ) has been hotly debated in the past@xcite . a possible _ caveat _ concerns the use of the egret data to derive the averaged gamma - ray spectra , since the limited sensitivity likely introduces a bias toward flaring states .
one argument of the critics is the claimed existence of `` blue quasars '' , fsrqs with peaks at high frequencies , violating the sequence .
however , so far none of the blue quasars candidates has been proven to be a robust case@xcite@xcite .
the three - month lat data confirm with a larger statistics@xcite one of the basic expectations of the blazar sequence , namely the fact that fsrqs should have ( on average ) soft gev spectra ( with energy index @xmath5 ) whereas bl lac should show hard spectra ( @xmath6 ) .
as shown in fig.[f1 ] this prediction is confirmed by the new data , showing not only the expected dicotomy in the spectral slope , but also a well definite trend between the gamma - ray luminosity and the slope of the gamma - ray spectrum . the small `` thickness '' ( or even the very existence ) of the correlation
is even surprising , considering the huge variability possible for both quantities in the same source .
this is shown for some selected objects by the solid lines line connecting different emission states of the same source . in ref .
we proposed that the divide in the gamma - ray luminosity ( a good proxy for the bolometric output of the jet ) can be associated to a transition in the properties of the accretion flow on the central bh , transiting from a standard shakura - sunyaev disk at high luminosities , to an inefficient accretion regime ( adaf@xcite , adios@xcite ) .
the transition occurs at the level of an accretion rate around @xmath7 in eddington units , in agreement with the expectation from the theory@xcite and the analogous scenario proposed to account for the fri / frii division among radiogalaxies@xcite .
the existence and the characteristics of the blazar sequence call for a robust and simple explanation .
interestingly , the spectral sequence is associated to a ( even more robust ) sequence in some of the fundamental quantities of the emission region , as modeled with one - zone leptonic models@xcite .
in particular , the energy of the electrons emitting at the peaks of the seds shows a strong correlation with the energy density ( radiative plus magnetic ) in the source .
the most direct interpretation is that in sources with a large energy density the severe radiative cooling prevents the electrons to reach high energy , determining a low energy frequency of the peaks in the seds . in sources
in which electrons are characterized small radiative losses the acceleration mechanism can instead bring electrons to higher energies , determining large peak frequencies .
electrons in the jet of fsrqs are commonly assumed to have a rather high cooling rate , since the environment external to the jet is expected to be rather rich of soft photons , determining severe radiative losses through compton emission .
bl lac objects , instead , are characterized by an environment poor of radiation ( and also possibly by a radiatively inefficient accretion flow ) and thus electrons can be energized to very high energy , the cooling being dominated by synchrotron and synchrotron - self compton losses .
we refer to ref . for a deeper discussion .
as discussed above , jets in bl lac objects are expected to propagate in a `` clean '' environment and the emission we observed from them is likely due to synchrotron and ssc mechanisms .
being less prone to depend on the details specifying the location of the emission zone ( e.g. l. stawarz and l. costamante , these proceedings ) and the conditions of the surrounding environment , the emission from bl lac objects seems to be the ideal tool to investigate the physics of the particle acceleration and cooling processes acting in jets .
moreover , bl lac objects are interesting sources since the high energy component the sed is located at the crossroad between the band covered by lat ( @xmath8 gev ) and by ground based cherenkov telescopes ( @xmath9 gev ) .
it is thus possible to probe the entire high - energy emission , fully covering the bump before and after the ic peak ( e.g. ref . ) . to investigate the global properties of the bl lac objects revealed at high and very high energies we considered@xcite all the bl lac objects in the lbas list , including also 12 sources detected at tev energies . to this sample
we add the remaining 12 bl lac detected at tev energies ( before the end of 2009 september ) not detected by lat .
for all these sources we assembled the sed using the high energy data , historical data and , when available the ( often nearly simultaneous ) optical - uv and x - ray data from the _ swift _ satellite .
we reproduced the sed with a simple one - zone synchrotron - ssc model , deriving some of the basic physical quantities . for simplicity
, the model assumes a spherical homogeneous region and the relativistic electrons are assumed to follow a broken power law distribution in energy with the break at the lorentz factor @xmath10 . in fig.[f2 ] we report the values of the magnetic field @xmath11 versus @xmath10 ( corresponding to the energy of the electrons emitting most of the power , close to the sed peaks ) and the doppler relativistic factor , @xmath12 .
most of the sources occupy a region specified by a magnetic field in the range @xmath13 g , @xmath14 , @xmath15 .
however , there are some notable exceptions , requiring very small magnetic fields ( @xmath16 ) g , large @xmath10 ( @xmath17 ) and/or @xmath18 . at a closer look these `` outliers '' can be divided into two classes : i ) sources showing a moderately hard lat spectrum with the synchrotron peak in the optical - uv band ; ii ) tev bl lacs not detected by lat characterized by extremely hard ( de - absorbed ) tev spectra .
the sed of one of the bl lacs belonging to the group i ) , pks 1717 + 177 , is given in fig.[f3 ] ( adapted from ref . ) .
besides the lat `` butterfly '' , calculated with the spectral data of the lbas catalogue@xcite , we report the datapoints derived by analyzing the lat data with the standard procedure .
the points confirm that the spectrum continues to be hard up to the highest energies
. for these kind of sources the large separation between the synchrotron ( optical - uv ) and the ssc ( @xmath9 gev ) peaks in the sed inevitably requires large doppler factors@xcite . in the case of pks 1717 + 177
we show the model resulting by using a moderate value of the doppler factor , @xmath19 ( solid line ) : the fit of the lat data is clearly poor . using a somewhat extreme value of @xmath20 (
dashed line ) clearly we obtain a better agreement with the lat data , although the lat spectrum would be better reproduced with a peak beyond 50 gev . with a word of caution due to the non simultaneity of the _ swift _ and lat data , we conclude that the simplest version of the ssc model would clearly be in trouble with these sources as far as we assume that @xmath12 can not exceed 50 . a viable solution to alleviate the problem represented by the large doppler factors is offered by the model assuming that the emission zone is not a single homogeneous region , but jet is structured , with a fast ( @xmath21 ) core ( or _ spine _ ) surrounded by a slower ( @xmath22 ) layer .
these models were originally introduced to account for the discrepancy between the large doppler factors inferred in some of the tev bl lacs and the slow speeds measured in their jets at vlbi scale@xcite and have been subsequently adopted to model the high energy emission of radiogalaxies@xcite .
the key ingredient is the radiative coupling between the spine and the layer : both components see the emission from the other relativistically boosted ( because of the relative speed ) .
the ic emission is therefore amplified with respect to the ssc case .
in particular , if the emission from the layer contains more soft photons than that of the spine , the corresponding ic - scattered photons can reach higher energy than the ssc photons , whose energy is strongly limited by the rapid decline of the klein - nishina cross - section . the emission from the spine could then show a high - energy tail from this kind of external compton component , accounting for the observed lat spectrum .
a full discussion , including more detailed models is in preparation . among tev
bl lacs there are a handful of peculiar sources showing very hard tev spectra ( when the observed spectrum is properly deabsorbed to take into account the absoprtion by photons of the cosmic ir - optical background ) . the sed of one of the most extreme cases , 1es 0229 + 200 , is reported in fig.[f4]@xcite . in the framework of the one - zone ssc model such hard spectra
are difficult to obtain , since the decreasing of the scattering efficiency in the klein - nishina regime leads to soft spectra . however , extremely hard spectra can be obtained@xcite if the emitting relativistic electrons have a large value of the minimum electron energy @xmath23 : in this case the synchrotron emission , below the typical frequency of the low - energy electrons , @xmath24 hz , will display the typical @xmath25 spectrum .
the _ same hard slope _ will be reflected by the ssc emission below the peak . in fig.[f4 ]
we report the optical - uv and x - ray data obtained by _
swift _ , showing that a hard spectrum below the soft x - ray band is required by the data , supporting the possibility discussed above .
clearly , another strong prediction of this scenario is that the level of the emission in the gev band should be rather low .
consistently , all the sources featuring such hard tev spectra have not been detected in the first months of fermi . in a recent paper@xcite
, the lat collaboration reports a deep ( 5.5 months of observations ) upper limit for 1es 0229 + 200 ( that we plot in fig.[f4 ] , dashed lines ) , fully consistent with the theoretical expectations .
if our scenario is correct , these sources would be the most efficient cosmic accelerator known , able to push virtually all the electrons at very high energy .
\1 ) the `` blazar sequence '' is confirmed and even strengthened by the data from lat .
i would like to remark , however , that deeper data will start to probe a regime of luminosities populated be a mix of sources .
in particular , blazars with small black holes or with jets slightly misaligned will introduce some confusion , making the discussion more difficult .
\2 ) the one zone synchrotron - ssc model for bl lacs gives satisfactorily result for most of the sources when we use almost simultaneous data .
however , for a real test of the model it is important to obtain really simultaneous data .
\3 ) the bl lac mainly populating the lat sky are , obviously , those with a prominent gev emission .
however , there is an interesting population of bl lacs , characterized by very hard gev - tev spectra that are not easily detected by lat .
any discussion on the high - energy emission of bl lacs based only on lat would exclude these sources that could be the most efficient machines accelerating the relativistic electrons .
i am indebted with g. ghisellini , l. maraschi , g. ghirlanda and l. foschini for years of fruitful collaboration .
i thank the organizers for the very fruitful and pleasant meeting .
this work was partly financially supported by a 2007 cofin - miur grant and by asi under contract asi i/088/06/. | blazars , radio - loud active galactic nuclei with the relativistic jet closely aligned with the line of sight , dominate the extragalactic sky observed at gamma - ray energies , above 100 mev .
we discuss some of the emission properties of these sources , focusing in particular on the `` blazar sequence '' and the interpretative models of the high energy emission of bl lac objects . |
the study of chromospheric and coronal activity has progressed impressively in the last two decades , thanks to the launch of x - ray satellites .
coronal sources have been detected in x rays all over the cool part of the h - r diagram ( vaiana et al .
1980 ) and active binaries of the rs cvn type have soon been recognized as powerful x - ray emitters ( walter et al .
1980 ) . with the advent of the rosat satellite and its all - sky survey ,
a complete x - ray study of known active binaries has been possible ( dempsey et al .
1993a , b ) .
a large number of previously unknown coronal active stars is becoming available ( see e.g. metanomski et al .
1998 ) , with interesting consequences on our understanding of the population of young stars in the solar neighborhood ( guillout et al . 1998 ) . while the general framework of dependence of x - ray emission on stellar rotational velocity ( and then stellar age ) has been established long time ago ( pallavicini et al . 1981 ) ,
there still remain many uncertainties about which stellar parameters affect the coronal emission in late - type stars . we know , for instance , that old stars may preserve a high level of x - ray emission if they are in binary systems , where a high rotational velocity can be maintained through tidal interaction , but it is not clear which role stellar mass , radius , orbital period and eccentricity play in determining the level of activity .
one of the most relevant limitations is that , since most studies are performed in field samples , it is difficult to determine precisely the characteristics of the studied objects . stars with different ( but poorly determined ) masses , ages and possibly evolutionary histories are often compared . the study of clusters , where ages , masses , and evolutionary status of the counterparts of x - ray sources can be well established , is a necessary step forward . in this framework ,
not only rosat observations of young clusters have provided new insight to the discussion on the age - activity relationship ( see e.g. randich and schmitt 1995 ) , but for the first time it has also been possible to detect coronal sources in old clusters , i.e. with ages comparable with the sun . with the rosat observations of m 67 and other intermediate age clusters , belloni et al.(1993 , 1996 , 1997 ) allowed to extend the study of the evolution of coronal activity to systems with age up to @xmath0 6 gyrs . by examining these coeval samples
, these observations allow the study of active coronae from a different , unique perspective : it is possible to investigate which stars ( or stellar systems ) show the highest activity level in samples so old that the emission from single , solar - type stars is expected to be very low .
the rosat observations require a follow up at different wavelengths , in order to : * confirm the identification of the optical counterparts . in these clusters
the stellar density is rather high , and more than one source may be contained in the rosat error box . as an example of the relevance of this issue
, the reader can compare the results of the present work ( summarized in table 1 ) with those of belloni , verbunt & schmitt ( 1993 ) , based only on x - ray and general optical information , but not on a detailed follow up .
* study in more detail the characteristics of the sources , their binary nature , x - ray and chromospheric emission . *
compare the observed characteristics with those of field stars . in this work ,
we present the identification and the study of the rosat counterparts of the old open cluster m 67 ( belloni , verbunt & schmitt 1993 ) . a second , longer rosat pointing to m
67 has been performed ( belloni , verbunt & mathieu in preparation ) : where possible , the results from this observation have been included in the present paper .
m 67 is a very interesting target , not only because its age and metallicity are similar to those of the sun , but also because the cluster has been subject to several photometric and spectroscopic studies which have led to detailed membership determinations ( sanders 1977 , girard et al . 1989 ) , a large amount of data on binary stars ( latham et al . 1992 ) , photometric variables , w uma candidates ( gilliland et al . 1991 ) , and blue stragglers ( mathys 1991 ) .
this large body of ancillary data makes m 67 one of the best studied clusters ; one am her system and a hot white dwarf belonging to the cluster have already been confirmed within this identification project ( pasquini et al .
the observations were carried out at eso , la silla , over the period 1992 - 1995 , using a variety of instruments and telescopes .
we stress that , being mainly interested in the coronal counterparts of rosat detections , we did not attempt a complete identification of all the sources , but only of the possible coronal cluster members .
this implies that , when only optically faint candidates are within the rosat error box , observations were not always pursued further .
first , observations were carried out at the eso 1.52 m telescope equipped with the b&ch spectrograph ( turatto 1997 ) : with a resolution of 2 / pixel and a 2048 pixel ccd , the range 3700 - 7600 was covered .
all objects within the reach of the telescope - spectrograph combination were observed , within @xmath0 40@xmath1 from the nominal rosat source in order to be confident that no possible counterpart would be missed .
a posteriori , we found this radius exceedingly large and we could confirm the good accuracy of the rosat error box .
bright stars within fields with no obvious counterpart have been observed even if located at a comparatively large distance from the x - ray position .
the low - resolution spectra were inspected to derive ( or check ) spectral types or spectral anomalies , but mostly to find signatures of high chromospheric emission , like filling - in or emission of the balmer lines .
most of the stellar candidates were selected in this way .
additional observations at low resolution needed for the identification of some of the fainter counterparts were performed using the 3.6 m telescope with the efosc spectrograph ( benetti et al .
1997 ; see also pasquini et al .
1994a ) .
finally , the 3.6 m telescope with the caspec spectrograph ( randich and pasquini 1996 ) and the ntt with the emmi spectrograph were used to obtain intermediate- and high - resolution spectroscopy of the pre - selected candidates .
this last step is required to allow a firm identification of the targets and to derive absolute chromospheric fluxes at the stellar surface .
the observations were centered in the ca ii h and k region , and in the h@xmath2 region .
caspec spectra have a resolving power r=18000 .
the emmi spectra were obtained in dichroic mode : blue and red spectra were recorded simultaneously .
blue spectra were acquired with a holographic grating at r=6000 ( pasquini et al .
1994b ) , while red spectra were obtained with a resolution r=20000 or r=3000 , depending on the stellar apparent magnitude .
the spectra have been reduced using the midas package ( banse et al .
1988 ) . when a candidate showed enhanced chromospheric activity and acceptable positional coincidence , it was accepted as counterpart .
however , for some of the fields no star fulfilled both conditions . for these fields , we obtained low resolution spectroscopy of fainter candidates , and when no acceptable alternatives were found , the original stars were accepted as ( possible ) candidates .
belloni , verbunt & schmitt ( 1993 , hereafter bvs ) list 22 sources detected in the central 20 arcmin of a rosat pspc field centered on a position @xmath3 east of the center of m 67 .
the rosat sources with firm or possible optical counterparts are summarized in table 1 : the numbering scheme is the same as in bvs .
source number 11 was recognized by bvs as a blend of a hard and a soft source . in the analysis of the new rosat observation of m 67 ( belloni , verbunt & mathieu , in preparation ) ,
the two sources could be resolved and they are listed separately ( sources 11a and 11b ) .
the optical positions are from girard et al .
( 1989 ) when available , while the coordinates for the remaining objects are from the digitized sky survey .
for these , distances are accurate only to a few arcsecs . for optical identification
, we adopted the results obtained by similar studies in the field ( see e.g. stocke et al
qso s and emission - line galaxies , when close to the nominal error box were accepted as counterparts ; in particular , in the error boxes of the 3 extragalactic sources identified with x - ray sources , no known m 67 member exists .
for the ten rosat fields with no identification , efosc images were acquired , but no stellar counterparts was found down to magnitude limits in excess of those expected for values of l@xmath4/l@xmath5 typical for coronal sources . in figure 1 , low resolution spectra of the extragalactic sources and of some of the most interesting objects in the fields with doubtful or no optical identification are shown .
note that the shape of the continuum may not represent the true continuum of the objects , since in order to gain spectra of several objects simultaneously , in most cases the slit could not be aligned with the parallactic angle .
thanks to the boresight correction applied by bvs , the optical and x - ray positions for most candidates agree extremely well , with differences often smaller than 10 arcsec , which is the typical nominal rosat error box ( bvs ) .
therefore , we consider the 2 proposed counterparts having larger distances from the x - ray source ( b1 , b15 ) only as possible , but not likely candidates . for b19 , the uncertainty in the identification is given by the fact that , although the x - ray and the optical position of the star s364 match quite well , the high - resolution spectra obtained do not show any sign of enhanced chromospheric activity ( cfr .
section 5.2 ) .
this , according to the previously mentioned criteria , does not make this star a firm counterpart .
0.2 cm 0.2 cm [ cols= " > , > , < , < , < " , ]
the aim of this work is to discuss the nature of the coronal sources belonging to m 67 detected in the rosat observation of bvs . for this purpose
we concentrate mostly on the firmly identified sources .
it is important to remember that the observations by bvs could only detect the high x - ray luminosity tail of the cluster .
the longer , more recent rosat observations reveal more sources , confirming that cluster x - ray emitters might exist at lower levels than the one analyzed here ( belloni , verbunt & mathieu , in preparation ) . of the 8 known members of table 2 , six are either known binaries or show clear signs of duplicity , like photometric modulation or displaced position in the colour - magnitude diagram of the cluster . of the remaining two ,
s1077 is reported by bvs as a multiple system with short period , while s364 is not included in the current list of binaries .
jones and smith ( 1984 ) found anomalous ddo colours for this star , but they concluded that this was probably due to due measurement uncertainties .
once again , we stress that this identification is only considered as ` possible ' . _
the first conclusion is that in m 67 the strongest coronal sources are binaries_. this was expected from the results on field stars and from the age - coronal activity relationships : single , late type stars with ages comparable to the sun are not expected to emit x - ray in excess to @xmath010@xmath6 erg / s in luminosity . in figure 3 the colour - magnitude diagram of the cluster
is shown , with the x - ray identifications marked .
it is striking to notice that a large variety of cases exists : one blue straggler ( s1082 ) , one peculiar object ( s1063 ) , evolved ` normal ' binaries ( s1077 , s999 ) , main sequence binaries ( s1019 , s972 ) , one red straggler ( s1040 ) , and possibly a red giant ( s364 ) . among field stars
such a comparison can hardly be done , due to the uncertainty in the fundamental stellar parameters . _
the second conclusion is therefore that among the m 67 sources there exists a large variety in evolutionary status and composition of binary systems . in particular ,
several of the strong x - ray emitters are found among objects having peculiar location in the colour magnitude diagram . _
this variety makes it difficult to understand which parameters determine the x - ray emission in these binaries . from the study of field active binaries
, it emerged that the main parameter determining x - ray luminosity is stellar radius .
therefore , the strongest x - ray emitters would be expected to be among the most luminous systems ( dempsey et al .
1993b ) . on the contrary , the most luminous x - ray sources in m 67 span over a range in visible luminosity of at least 4 magnitudes ( cfr
. table 1 ) .
although s972 , which has the lowest x - ray luminosity , is also the faintest established member of the cluster in the optical , other stars ( like s1019 or s1063 ) show the highest x - ray emission , while being relatively faint in the optical .
another interesting point is the dependence of x - ray emission on the orbital ( or rotational ) period , since x - ray luminosity scales with stellar rotational velocity ( pallavicini et al .
1981 ) , which in turn is related to stellar radius and orbital period . in most cases ( and for main sequence stars )
, it is expected that short period binaries are synchronized , i.e. they have equal orbital and rotational periods .
it is also expected that synchronization happens before orbit circularization ( zahn 1977 ) .
all 4 stars for which the rotational modulation is known , as inferred from the photometric variability , have rotational periods of less than 10 days . for s1063
, only the orbital period is known , but this star may have a special history ( see section 5.3 ) .
pending more data on the rotational period of the other objects , we can say that on the other hand , do all m 67 short period systems show strong x - ray ( or chromospheric ) activity ? in m 67 ,
11 binaries have a measured orbital period shorter than 16 days ( latham et al . 1992 ) , a canonical value for the definition of rs cvn systems , but most of them were not detected in the rosat observations .
three of them ( s1272 , s1284 , s1224 ) where out or at the very border of the rosat field ; one ( s999 ) was detected , but the remaining ( s1045 , s1234 , s1024 , s986 , s1009 , s1070 s1014 , s810 ) were contained in the rosat field but not detected .
these systems are likely active , but with x - ray luminosities l@xmath7 30.3 ( corresponding to the sensitivity of bsv ) , as it is suggested by the likely detection of some of these stars in the deeper pointing ( s1045 , s1024 , s1234 and s1070 , belloni , verbunt & mathieu , in preparation ) .
since some of the non - detected binaries , like s986 , s1009 and s810 , have periods of 10 days or shorter , the data collected up to now show that since we are dealing mostly with evolved stars , for which short periods can coexist with eccentrical orbits , eccentricity could be a relevant parameter , because the non - synchronization would allow the possibility of having a rotational period shorter than the orbital period .
of the detected binaries , s1063 and s999 have highly eccentric orbits , but s1040 has an eccentricity comparable with 0 .
this is surprising , because if the photometric variability detected by gilliland et al .
( 1991 ) really represents the rotational period , than this star would be circularized but not synchronized . on the other hand
, s1040 had probably a very complex history , having had mass transfer in the past ( landsman et al .
1997 ) . considering this star as a special case
we , could argue that non - circular binaries are favored among the strongest x - ray emitters . however , in the list of latham et al .
( 1992 ) , there are five short - period binaries with eccentricities significantly different from 0 ( s1284 , s1272 , s1234 , s1224 , s1014 ) .
two of them ( s1014 and s1234 ) were in the rosat field of view but were not detected .
therefore , it seems that also a high eccentricity does not represent _ per se _ a condition for strong x - ray emission . the chromospheric flux of the sun in the ca ii k line is of 3 - 5@xmath810@xmath9 erg @xmath10sec@xmath11 ( pasquini et al .
1988 ) . for the main sequence g stars in m 67 , being metallicity and age comparable with those of the sun
, values comparable with this are expected .
recently , dupree et al .
( in preparation ) studied the chromospheric emission in the ca ii lines for a sample of m 67 giants , using calibrations and spectra of similar quality to the ones presented here . in figure 4 the ca ii
chromospheric fluxes are plotted as a function of the ( v - r ) colour , both for the stars in dupree et al .
( open squares ) and for the x - ray counterparts of table 2 ( filled squares ) .
the rosat stars have a level of chromospheric activity one order of magnitude higher than the optically - selected giants in dupree et al .
the only exception is s364 , which fits extremely well within normal giants of similar spectral type .
for this reason , despite the reasonable agreement with the x - ray position , this star is considered as a doubtful counterpart . in order to perform an unbiased analysis of the relationship between chromospheric and coronal activity
, we have to use indicators which express similar quantities , namely fluxes at the stellar surface .
we computed the radii of the stars by using the same barnes - evans ( 1976 ) relationship used to calibrate the ca ii data : log@xmath12 = 0.4874 - 0.2v@xmath13 + 0.858(v - r ) .
although we know that the resulting radii are probably incorrect ( for instance , because the stars are implicitly assumed to be single ) , the fact that the same relationship is used for computing the ca ii and x - ray fluxes minimizes the presence of possible systematic effects in the comparison .
we assumed an absorption a(v)=0.17 and a distance of 785 parsecs ( janes 1984 ) .
the resulting diameters and radii are given in tables 2 and 3 ( in units of milliarcseconds and 10@xmath14 cm respectively ) . in figure 5a , log(f@xmath4 ) is given as a function of log f@xmath15 .
the relationship is rather scattered and mostly the presence of the ( doubtful ) s364 hints to the presence of a trend .
since the sample contains systems whose evolutionary status is very different from each other , to further investigate this point we plot in figure 5b the ratio between the x - ray and chromospheric fluxes versus apparent magnitude .
fainter ( i.e. higher gravity ) stars have much higher coronal to chromospheric flux ratios than more luminous ( i.e. lower gravity ) stars .
although the number of objects is rather low , it appears that dwarfs and giants follow different trends , with dwarfs having higher f@xmath4 for a given f@xmath15 .
this fact has two possible explanations : * for comparable chromospheric fluxes , higher - gravity stars are more efficient in heating their coronae than lower - gravity stars .
this could indicate the presence of different coronal structures between dwarfs and evolved stars . *
the assumption that the ca ii k fluxes are representative of the whole chromospheric losses may not apply when comparing stars of different luminosity .
giants could for instance have a different balance in the different chromospheric lines than dwarfs .
s1113 was was outside the field of the observation by bsv .
because of the strong chromospheric activity observed , we would expect it to show rather strong x - ray emission , which is indeed detected in the new rosat pointing ( cfr .
table 3 ) .
these two stars are located below the giants branch : they are as red as single subgiants , but almost one magnitude fainter .
s1063 is a known eccentric binary with a period of 18.3 days , while s1113 is a short period ( 2.82 d ) circular binary ( latham et al .
1992 , mathieu et al . in preparation ) .
the two stars are classified as members in the proper motion studies of sanders ( 1977 ) and girard et al .
( 1989 ) , with a probability higher than 90@xmath16 .
their peculiar position in the colour - magnitude diagram and their high level of coronal and chromospheric activity make these two object very interesting .
it is not possible to simply combine two m 67 stars and obtain the magnitudes and colours of s1063 and s1113 .
some mass exchange , or large mass losses in the past history of the systems , possibly still going on , look unavoidable . the high - resolution ca ii spectrum of s1113 shows neither direct evidence of duplicity , nor strong asymmetries in the ca ii core typical of strong mass losses , but
not much can be derived with only one optical spectrum for this star .
we stress that the conclusion that these two stars have suffered a special evolutionary history is made possible only by the fact that we they are members of a cluster ( and therefore we can firmly position them in the colour magnitude diagram ) and by the detailed optical follow - up .
how many such systems exist among field binaries ?
note that , in absence of detailed studies ( e.g. accurate determinations of mass , metallicity and gravity ) , similar systems in the field can not be distinguished by otherwise normal rs cvn binaries .
a more detailed study in other clusters and possibly the analysis of hipparcos parallaxes of active binaries will help in understanding how common these systems are .
investigation of their binarity and orbital synchronization will also be crucial to model their possible evolution with time ( see i.e. the discussion in stepien 1995 ) .
one of the aims of this study is the possibility of comparing for the first time active stars in an old cluster with the field population of rs cvn .
dempsey et al .
( 1993a , b ) studied the rosat detections of known rs cvn .
their sample , taken from the catalogue of strassmeier et al .
( 1988 ) , contains binaries of a large variety of ages , masses and periods .
we selected all the stars from dempsey et al .
( 1993a ) with known distance .
we computed x - ray luminosities by converting the pspc count rates ( see also dempsey et al .
1994 ) using the same conversion factor used for the m 67 stars .
the distribution in luminosity is given in figure 6 for m 67 and field rs cvn separately .
although the m 67 sources do overlap well with the main body of field rs cvn s , their emission is not as high as the most active rs cvn systems .
the reasons for this difference are at the moment not very clear : it could be an effect due to the old age of the cluster and/or to the evolutionary status of the sources , or to a statistical effect caused by the fact that we can only sample a few hundred stars within the cluster .
identifications in clusters of different ages , as well as a detailed analysis of the strongest sources in the field , will help in understanding this open question .
it is interesting to note that in the analysis of the much younger hyades , stern et al .
( 1995 ) found that binaries are the strongest x - ray emitters in this cluster .
similarities with m67 exist also in that the strongest x - ray source in the hyades ( v471 tau ) is a peculiar system .
finally , hyades binaries have x - ray luminosities similar or lower than those observed among the m67 sources .
this could indicate that as far as the high - luminosity tail of the x - ray luminosity function is concerned , age is not the primary parameter to determine the x - ray emission in binaries , as pointed out also by ottmann et al .
( 1997 ) in their analysis of pop ii binaries .
we would therefore argue that the extremely active i.e. ( l@xmath4 @xmath0 10@xmath17 erg @xmath10 sec @xmath11 ) rs cvn systems seem to be very rare and possibly limited to quite exceptional cases .
metanomski et al . (
1998 ) , in their study of hundred stars identified over a large area of the rosat all - sky survey , find results similar to ours : no such a high luminosity coronal source is indeed present in their sample .
banse , k. , grosbol , p. , ponz , d. et al 1988 ` the midas image processing system ' in ` instrumentation for ground base astronomy : present and future ' , robinson , l.b .
( springer ) barnes , t.g . , evans , d. 1976 , mnras 174 , 489 basri , g. 1987 , apj , 316 , 377 belloni , t. , verbunt , f. , schmitt , j.h.m.m .
1993 , a&a 269 , 165 ( bvs ) belloni , t. , verbunt , f. 1996 , a&a 305 , 806 belloni , t. , & tagliaferri , t. , 1997 , a&a , 326 , 608 benetti , s. , savaglio , s. pasquini , l. 1997 , ` efosc operating manual ' ( eso ) dempsey , r.c . ,
linsky , j.l . ,
schmitt , j.h.m.m .
, fleming , t.a .
1993b , apj , 413 , 333 dempsey , r.c . ,
linsky , j.l .
, fleming , t.a . ,
schmitt , j.h.m.m .
1993a , apjs , 86 , 599 dempsey , r.c . ,
linsky , j.l .
, fleming , t.a . , schmitt , j.h.m.m .
1994 , apjs , 94 , 829 fleming , t.a . ,
snowden , s.l .
, pfeffermann , e. , briel , u. , greiner , j. 1996 , a&a , 316 , 147 gilliland , r. et al .
1991 , aj , 101 , 541 girard , t.m . ,
grundy , w.m . , lopez , c.e .
, van altena , w.f .
1989 , aj 98 , 227 goranskij , v.p . ,
kusakin , a.v . ,
mironov , a.v . ,
moshkaljov , v.g . ,
pastukhova , e.n .
1992 , astron .
trans . 2 , 201 guillout , f. , sterzik , m.f . ,
schmitt , j.h.m.m .
1998 , a&a 332 , l29 hasinger , g. , burg , r. , giacconi , r. et al .
1998 , a&a 329 , 482 janes , k.a . 1984 , in calibration of fundamental stellar quantities , iau symposium no 11 , d.s .
hayes , l.e .
pasinetti , a.g .
davis philip eds .
( reidel ) janes , k.a . ,
smith , g.h .
1984 , aj 89 , 487 johnson , h.l . 1966 ,
ara&a 4 , 193 landsman , w. , aparicio , j. , bergeron , p. , di stefano , r. , stecher , t.p .
1997 , apj , 481 , l93 latham , d.w . ,
mathieu , r.d .
, milone , a.e . , davis , r.j .
1992 , in ` binaries as tracers of stellar formation ' , a. duquennoy and m. mayor eds .
( cambridge un .
mathys , g. 1991 , a&a 245 , 467 melnick , j. 1993 , emmi operating manual ( eso ) metanomski , a.d.f . , pasquini , l. , krautter , j. , cutispoto , g. , fleming , t.a .
1998 , a&as , in press montgomery , k.a . , marschall , l.a . , janes , k.a .
1993 aj , 106 , 181 ottmann , r. , fleming , t.a . ,
pasquini , l. 1997 , a&a 322 , 785 pallavicini , r. , golub , l. , rosner , r. , vaiana , g.s .
, ayres , t. , linsky , j.l .
1981 , apj 248 , 279 pasquini , l. , pallavicini , r. , pakull , m. 1988 , a&a 191 , 266 pasquini , l. , pallavicini , r. , dravins , d. 1989 , a&a 213 , 261 pasquini , l. , belloni , t. , abbott , t.m.c .
1994a , a&a 290 , l17 pasquini , l. , storm , j. , dekker , h. 1994b , the messenger 77 , 5 randich , s. , schmitt , j.h.m.m .
1995 , a&a , 298 , 115 randich , s. , pasquini , l. 1996 , caspec operating manual ( eso ) sanders , w.l .
1977 , a&as 27 , 89 stepien , k. 1995 , mnras 274 , 1019 stern , r.a . ,
schmitt , j.h.m.m .
, kahabka , p.t . , 1995 ,
apj 448 , 683 stocke , j.t . , morris , s.l . ,
1991 , apjs , 76 , 813 strassmeier , k.g . , hall , d.s . , zeilik , m. , nelson , e. , eker , z.,fekel , f.c . , 1988 , a&as , 72 , 291 turatto , m. 1997 , ` the b&ch spectrograph user s manual ' ( eso ) vaiana , g.s . ,
cassinelli , j.p .
1981 , apj 245 , 163 walter , f.m . ,
cash , w. , charles , p.a . , bowyer , c.s .
1980 , apj 245 , 671 zahn , j.p . , 1977 , a&a 57 , 383 | we present optical identification and high - resolution spectroscopy of rosat sources in the field of the old open cluster m 67 . for the first time it is possible to analyze coronal and chromospheric activity of active stars in a solar - age cluster , and to compare it with field stars .
rosat observed the high x - ray luminosity tail of the cluster sources . in agreement with what expected from studies of field stars ,
most of the detected x - ray sources are binaries , preferably with short periods and eccentric orbits .
in addition , several of the m 67 rosat sources have peculiar locations in the cluster colour - magnitude diagram .
this is most likely due to rather complex evolutionary histories , involving the presence of mass transfer or large mass losses .
the x - ray luminosity of the sources does not scale with the stellar parameters in an obvious way .
in particular , no relationship is found between coronal emission and stellar magnitude or binary period .
the ca ii k chromospheric flux from most of the counterparts is in excess to that of single stars in the cluster by one order of magnitude .
the x - ray luminosity of the sources in the old m 67 is one order of magnitude lower than the most active active binaries in the field , but comparable to that of the much younger binaries in the hyades . |
because of its intermediate mass , the strange quark has a special status in low - energy qcd . it is light enough to allow for a combined expansion of observables in powers of @xmath3 around the @xmath4 chiral limit ( meaning 3 massless flavours ) : @xmath5 .
but it is sufficiently heavy to induce significant changes from the @xmath4 limit to the @xmath6 limit : @xmath7 and @xmath0 physical .
each limit can engender its own version of chiral perturbation theory ( @xmath8pt ) . in the @xmath6 limit ,
the pions are the only degrees of freedom , whereas @xmath4 @xmath8pt deals with pions , kaons and @xmath9 .
this second version of @xmath8pt is richer , discusses more processes in a larger range of energy , but contains more unknown low - energy constants ( lecs ) and may have a slower convergence .
the details of the connection between the two theories remain under debate .
indeed , due to @xmath10 sea - pairs , order parameters such as the quark condensate and the pseudoscalar decay constant , @xmath11 and @xmath12 , can reach significantly different values in the two chiral limits ( @xmath13 denoting the chiral limit with @xmath14 massless flavours ) @xcite .
an illustation is provided by the quark condensate : @xmath15 here , @xmath10-pairs are involved through the two - point correlator @xmath16 , which violates the zweig rule in the vacuum ( scalar ) channel .
one expects @xcite that this effect should suppress order parameters when @xmath17 : @xmath18 and @xmath19 . since the quark condensate(s ) and the pseudoscalar decay constant(s ) are the leading - order lecs for the two versions of @xmath8pt , a strong decrease from @xmath6 to @xmath20 would have an impact on the structure of the two theories .
a few years ago , the e865 collaboration provided new data on @xmath21 decays @xcite . building upon the dispersive analysis of @xmath1 scattering @xcite , we extracted the two - flavour order parameters @xcite : @xmath22 a different analysis , with an additional theoretical input from the scalar radius of the pion , led to an even larger value for @xmath23 @xcite . the situation is somewhat modified by new data from the na48 collaboration @xcite , which show some discrepancy with the e865 phase shifts in the higher end of the allowed phase space .
the role of isospin breaking corrections is under discussion currrently .
the preliminary values of the phase shifts @xcite tend to increase the value of the @xmath24 @xmath1 scattering length , and to decrease the value of the two - flavour quark condensate , pushing @xmath23 down to 0.7 . in any case
, one would expect values closer to 1 , since @xmath23 and @xmath25 monitor the convergence of @xmath6 chiral expansions of @xmath26 and @xmath27 respectively .
such expansions in powers of @xmath28 and @xmath29 _ only _ should exhibit smaller nlo corrections ( below 10% ) @xcite . to include @xmath30- and @xmath9-mesons dynamically
, one must use three - flavour @xmath8pt around the @xmath4 chiral limit .
strange sea - quark loops may affect chiral series by damping the leading - order ( lo ) term , which depends on @xmath31 and @xmath32 , and by enhancing next - to - leading - order ( nlo ) corrections , in particular when violating the zweig rule in the scalar sector .
take for instance : @xmath33 where @xmath34 , and we have put together nlo low - energy constants and chiral logarithms @xmath35 , @xmath36 ( enhanced by @xmath0 ) .
if we assume that the lo contribution is numerically dominant ( i.e. , @xmath37 to a very good approximation ) , we can perform the following manipulations : @xmath38 where we have used @xmath39 and eq .
( [ eq : fpi ] ) at the second step , and the second and third terms of the last equality are obtained using @xmath40 and @xmath41 respectively @xcite .
this is clearly in contradiction with the orginal assumption @xmath42 .
the available dispersive estimates of @xmath43 @xcite yield a similar situation for @xmath44 citeresum .
therefore , a small positive value of @xmath45 or @xmath46 is enough to spoil the rapid convergence of @xmath4 chiral series and to induce a numerical competition between formal lo and nlo contributions .
( left ) and @xmath47 ( right ) . in each case
, the results are obtained from @xmath1 scattering only ( dashed line ) , @xmath2 scattering only ( dotted line ) , or both sources of information ( solid line).,title="fig:",height=211 ] ( left ) and @xmath47 ( right ) . in each case , the results are obtained from @xmath1 scattering only ( dashed line ) , @xmath2 scattering only ( dotted line ) , or both sources of information ( solid line).,title="fig:",height=211 ]
the potentially `` large '' values of @xmath48 and @xmath43 lead to a numerical competition between formal lo and nlo contributions in chiral series . to deal with such a situation ,
we have introduced a framework , called resummed chiral perturbation theory ( re@xmath8pt ) @xcite , where we define the appropriate observables to consider and the treatment of their chiral expansion @xcite .
it allows for a a resummation of the potentially large effect of the zweig - rule violating couplings @xmath48 and @xmath43 @xcite .
since this framework copes with the possibility of a numerical competition between ( formal ) lo and nlo terms in chiral series , some usual @xmath49 results are not valid any longer : for instance @xmath50 is not fixed by @xmath51 and may vary from 8 to 40 . one can apply this framework to @xmath1 and @xmath2 scatterings , which provide information on @xmath6 and @xmath4 patterns of chiral symmetry breaking respectively , and in particular on @xmath50 , the quark condensate @xmath47 and the decay constant @xmath52 .
one can exploit dispersive relations , such as roy equations @xcite and roy - steiner equations @xcite , to reconstruct the amplitudes from the phase shifts from threshold up to energies around 1 gev .
matching the dispersive and chiral representations of the amplitude in a frequentist framework provides constraints ( in terms of confidence levels ) on the main parameters of interest for three - flavour @xmath8pt @xcite .
[ fig : cls ] shows the situation for @xmath50 and @xmath47 .
the main impact of @xmath1 scattering consists in constraining @xmath50 : indeed , @xmath1 scattering pins down the two - flavour condensate @xmath23 rather accurately , which can be related to @xmath53 through the spectrum of pseudoscalar mesons @xcite .
the combination of @xmath1 and @xmath2 scatterings yields : @xmath54\ ] ]
the presence of massive @xmath55-pairs in the qcd vacuum may induce significant differences in the pattern of chiral symmetry breaking between the @xmath6 and @xmath4 chiral limits .
this effect , related to the violation of the zweig rule in the scalar sector , may spoil the convergence of three - flavour chiral expansions .
we introduce resummed chiral perturbation theory to deal with such a problem , and apply it to our experimental knowledge on @xmath1 and @xmath2 scatterings .
the outcome does not favour the usual picture of a large quark condensation independent of the number of massless flavours .
further experimental information is needed to constrain the pattern of chiral symmetry breaking efficiently and learn more about its variation with @xmath14 .
s. descotes - genon , l. girlanda and j. stern , jhep 0001 ( 2000 ) 041
. s. pislak _ et al .
_ , phys .
d * 67 * ( 2003 ) 072004 .
b. ananthanarayan , g. colangelo , j. gasser and h. leutwyler , phys .
* 353 * ( 2001 ) 207 .
s. descotes - genon , n. fuchs , l. girlanda and j. stern , eur .
j. c 24 ( 2002 ) 469 .
g. colangelo , j. gasser and h. leutwyler , phys .
lett . 86 ( 2001 ) 5008 .
g. lamanna , _
@xmath1 scattering from @xmath56 decays _ , talk at this conference .
p. buettiker , s. descotes - genon and b. moussallam , eur .
j. c 33 ( 2004 ) 409 .
s. descotes - genon , n. fuchs , l. girlanda and j. stern , eur .
j. c 34 ( 2004 ) 201 .
b. moussallam , eur .
j. c 14 ( 2000 ) 111 ; jhep 0008 ( 2000 ) 005 .
s. descotes - genon , jhep 0103 ( 2001 ) 002 .
s. descotes - genon , accepted for publication in eur .
j. c , arxiv : hep - ph/0703154 . | the ( light but not - so - light ) strange quark may play a special role in the low - energy dynamics of qcd .
the presence of strange quark pairs in the sea may have a significant impact of the pattern of chiral symmetry breaking : in particular large differences can occur between the chiral limits of two and three massless flavours ( i.e. , whether @xmath0 is kept at its physical value or sent to zero ) .
this may induce problems of convergence in three - flavour chiral expansions . to cope with such difficulties , we introduce a new framework , called resummed chiral perturbation theory .
we exploit it to analyse @xmath1 and @xmath2 scatterings and match them with dispersive results in a frequentist framework .
constraints on three - flavour chiral order parameters are derived . |
one of the classical problems of statistical physics , and physical chemistry , is to find a macroscopic statistical description for the diffusion of a dissolved molecule or ion in a uniform fluid solvent .
the microscopic state of such a system has very many degrees of freedom , possibly even more than an avogadro number of degrees of freedom .
this gives rise to an equally large number of coupled equations of motion .
it is completely impractical to solve such a large system with rigor .
we must abandon the idea of an exact solution .
we are forced to use a statistical description , where we can only describe the probability of certain events .
we denote the probability of finding a brownian particle at a certain point on space , @xmath0 , and time , @xmath1 , by @xmath2 .
the time - evolution of @xmath3 is governed by a partial differential equation called the fokker - planck equation : @xmath4 the functions @xmath5 and @xmath6 are referred to as the infinitesimal first and second moments of diffusion . in practice ,
the infinitesimal second moment does sometimes depend on concentration of the solute , @xmath3 , but is usually regarded as constant and is called the `` fick s law constant . '' a typical value ( for a hydrated sodium ion in water ) would be of the order @xmath7 .
the infinitesimal first moment depends on the magnitude of externally imposed forces and on the mobility of the brownian particle which is given by @xmath8 where @xmath9 is the electrical charge on the particle , @xmath10 is the kinematic viscosity of the solvent and @xmath11 is the effective radius of the particle .
a typical value for the mobility ( of a hydrated sodium ion in water ) would be @xmath12 .
further descriptions and numerical data may be found in books on physical chemistry and statistical physics @xcite .
if we apply an electrical potential , or voltage , of @xmath13 then the infinitesimal first moment is given by @xmath14 the theory behind equations [ eq : mobility ] and [ eq : infinitesimal_first_moment ] is due to stokes and einstein @xcite .
more information about the methods of solution and the applications of the fokker - planck equation can be found in risken @xcite .
when we take into account the functional forms of @xmath15 and @xmath16 then we can rewrite the fokker - planck equation as : @xmath17 this is the form of the fokker - plank equation which we will sample at regular intervals in time and space , to yield finite difference equations .
many partial differential equations , or pdes , including equation [ eq : expanded_fokker_planck ] , can be very difficult to solve analytically .
one well established approach to this problem is to sample possible solutions to a pde at regular intervals , called mesh points @xcite .
the true solution is approximated locally by a collocating polynomial .
the values of the derivatives of the true solution are approximated by the corresponding derivatives of the collocating polynomial .
we can define local coordinates , expanded locally about a point @xmath18 we can map points between a real space @xmath19 and an integer or discrete space @xmath20 .
time , @xmath1 , and position , @xmath0 , are modelled by real numbers , @xmath21 and the corresponding sampled position , @xmath22 , and sampled time , @xmath23 , are modelled by integers @xmath24 .
we sample the space using a simple linear relationship @xmath25 where @xmath26 is the sampling length and @xmath27 is the sampling time . in order to map equation [ eq : expanded_fokker_planck ] into discrete space
, we need to make suitable finite difference approximations to the partial derivatives .
the notation is greatly simplified if we define a family of difference operators : @xmath28 in principle , this is a doubly infinite family of operators but in practice we only use a small finite subset of these operators .
this is determined by our choice of sampling points .
this choice is not unique and is not trivial .
the set of sampling points is called a `` computational molecule @xcite . ''
some choices lead to over - determined sets of equation with no solution . some other choices
lead to under - determined sets of equations with infinitely many solutions .
we chose a computational molecule called `` explicit '' computation with the following sample points : @xmath29 .
we also need to make a choice regarding the form of the local collocating polynomial .
this is not unique and inappropriate choices do not lead to unique solutions .
a polynomial which is quadratic in @xmath0 and linear in @xmath1 is the simplest feasible choice : @xmath30 where @xmath31 , @xmath32 and @xmath33 are the real coefficients of the polynomial .
equations [ eq : sample ] , [ eq : difference ] and [ eq : polynomial ] imply a simple system of linear equations that can be expressed in matrix form : @xmath34 \left [ \begin{array}{c } a_1 \\ a_2 \\ b_1 \end{array } \right ] = \left [ \begin{array}{c } \delta_{-1,-1 } \\
\delta_{0,-1 } \\
\delta_{+1,-1 } \end{array } \right]~. \label{eq : finite_difference_conversion}\ ] ] these can be solved algebraically , using cramer s method to obtain expressions for @xmath31 , @xmath32 and @xmath33 : @xmath35 and @xmath36 and @xmath37 these are all intuitively reasonable approximations but their choice is not arbitrary .
equations [ eq : a_1 ] , [ eq : a_2 ] , [ eq : b_1 ] form a complete and consistent set .
we could not change one without adjusting the others .
we can evaluate the derivatives of equation [ eq : polynomial ] to obtain a complete and consistent set of finite difference approximations for the partial derivatives : @xmath38 and @xmath39 and @xmath40 we can apply the same procedure to @xmath5 to obtain @xmath41 equations [ eq : dpdz ] , [ eq : d2pdz2 ] , [ eq : dpdt ] and [ eq : dalphadz ] can be substituted into equation [ eq : expanded_fokker_planck ] to yield the required finite partial difference equation : @xmath42 where @xmath43 and @xmath44 and @xmath45 we can overload the arguments of @xmath46 and write them in terms of the discrete space @xmath20 using the mapping defined in equation [ eq : sample ] . @xmath47 the meaning of the arguments should be clear from the context and from the use of subscript notation , @xmath48 , rather than function notation , @xmath49 .
equation [ eq : integer_partial_difference_equation ] is precisely the form required for parrondo s games .
in the original formulation , the conditional probabilities of winning or losing depend on the state , @xmath22 , of capital but not on any other information about the past history of the games : * game a is a toss of a biased coin : @xmath50 where @xmath51 is an adverse external bias that the game has to `` overcome '' .
this bias , @xmath51 , is typically a small number such as @xmath52 , for example @xcite . * game b depends on the capital , @xmath22 : + if @xmath53 : : , then the odds are unfavorable .
@xmath54 if @xmath55 : : , then the odds are favorable .
@xmath56 it is straightforward to simulate a randomized sequence of these games on a computer using a very simple algorithm @xcite .
we can write the requirements for game a in the form of equation [ eq : integer_partial_difference_equation ] .
@xmath57 this implies a constraint that @xmath58 which implies that @xmath59 which defines the relative scales of @xmath26 and @xmath27 so we can give it a special name : @xmath60 the constraints on @xmath61 and @xmath62 imply a value for parrondo s `` @xmath51 '' parameter : @xmath63 which can be related back to an externally imposed electric field , @xmath64 using equations [ eq : mobility ] and [ eq : infinitesimal_first_moment ] : @xmath65 the small bias , @xmath51 , is proportional to the applied external field which justifies parrondo s original intuition .
there is still zero probability of remaining in the same state which implies a constraint that @xmath58 which implies that we still have the same scale , @xmath66 .
if we are in state @xmath22 then we can denote the probability of winning by + @xmath67 .
we can write the difference equations for game b in the form : @xmath68 which , together with equations [ eq : am1 ] , [ eq : a0 ] and [ eq : ap1 ] , gives @xmath69 which implies that @xmath70 this can be combined with equation [ eq : infinitesimal_first_moment ] and then directly integrated to calculate the required voltage profile .
we can approximate the integral with a riemann sum : @xmath71 so we can construct the required voltage profile for the ratchet which means that , given the values of @xmath72 , it is possible to construct a physical brownian ratchet that has a finite difference approximation which is identical with parrondo s games .
we can conclude that parrondo s games are literally a finite element model of a flashing brownian ratchet .
we note that game b , as defined here , is quite general and actually includes game a as a special case .
we would like to think that as long as @xmath73 is preserved then the solution to the finite partial difference equation [ eq : integer_partial_difference_equation ] would converge to the true solution of the partial differential equation [ eq : expanded_fokker_planck ] , as the mesh size , @xmath26 goes to zero .
fortunately , there is a theorem due to obrien , hyman and kaplan @xcite which establishes that the numerical integration of a parabolic pde , in explicit form , will converge to the correct solution as @xmath74 and @xmath75 provided @xmath76 .
similar results may also be found in standard texts on numerical analysis @xcite .
we see that parrondo s choice of diffusion operator , with @xmath66 is at the very edge of the stable region .
there is a possible range of values for @xmath77 .
as @xmath78 we require the time step @xmath75 which means that the number of time steps required to simulate a given time interval , @xmath79 , increases without bound @xmath80 .
it is computationally infeasible to perform simulations with very small values of @xmath77 . on the other hand ,
the value of @xmath81 implied in parrondo s games is at the very limit of stability .
in fact , the presence of small roundoff errors in the arithmetic could cause the the discrete simulation to diverge significantly from the continuous solution .
we propose that choosing @xmath82 , in the middle of the feasible range , is most appropriate .
if we consider the case of pure diffusion , with @xmath83 , then equation [ eq : integer_partial_difference_equation ] reduces to @xmath84 and if we choose @xmath82 then this reduces to @xmath85 which is the same as pascal s triangle with every second row removed .
the solution to the case where the initial condition is a kronecker delta function , @xmath86 is easy to calculate : @xmath87 which is a half period , or double frequency , binomial .
we can invoke the laplace and de moivre form of the central limit theorem which establishes a correspondence between binomial ( or bernoulli ) distribution and the gaussian distribution to obtain @xmath88 this expression is only approximate but is true in the limiting case as @xmath89 . in the case where @xmath90 ; the fokker planck equation [ eq : expanded_fokker_planck ] reduces to a diffusion equation : @xmath91 einstein s solution to the diffusion equation is a gaussian probability density function : @xmath92 where the variance , @xmath93 , is a linear function of time : @xmath94 it is possible to verify that this is a solution by direct substitution : @xmath95 if we sample this solution in equation [ eq : gaussian ] using the mapping in equation [ eq : sample ] then we obtain equation [ eq : laplace_demoivre ] again .
this is an exact result .
we conclude that the choice of @xmath96 is very appropriate for the solution to the diffusion equation .
we suggest that this would also be true for the fokker - planck equation , in the case where @xmath16 is `` small . ''
the appropriate choice of @xmath77 , given arbitrarily large , or rapidly varying , @xmath16 is still an unsolved problem . in general
, we would expect that much smaller values , @xmath97 , would be needed to accommodate more extreme choices of alpha .
we simulated a physically reasonable ratchet with a moderately large modulo value , @xmath98 .
( the value for the original parrondo s games was @xmath99 . ) we used the value of @xmath82 .
the simulation was based on a direct implementation of equation [ eq : integer_partial_difference_equation ] in matlab .
we chose a sampling time of @xmath100 and a sampling distance of @xmath101 .
the result is shown in figure [ fig : simulation ] , where we indicate how the expected position of a particle can move within a brownian flashing ratchet during four cycles of the modulating field .
we can see a steady drift of the mean position of the particle in response to the ratchet action .
this simulation includes a total of 500 time samples .
note that the average rate of transport quickly settles down to a steady value , even after only four cycles of the ratchet .
we acknowledge the similar , but independent , work of heath @xcite et al .
the focus of our paper is different . we seek to establish the physical , and mathematical , basis of parrondo s games and to derive a practical numerical technique for simulation .
we conclude that parrondo s games _ are _ a valid finite - element simulation of a flashing brownian ratchet , which justifies parrondo s original intuition . we have established that parrondo s `` @xmath51 '' parameter is a reasonable way to simulate a gradual externally imposed electric field , or voltage gradient .
we have established that parrondo s implied choice of the @xmath77 parameter does lead to a stable simulation but we suggest that the choice of @xmath82 is more appropriate from a mathematical point of view . finally , we have generalised parrondo s games , in the form of a set of finite difference equations [ eq : integer_partial_difference_equation ] and we have shown that these can be implemented on a computer .
j. s. bader , r. w. hammond , s. a. henk , m. w. deem , g. a. mcdermott , j. m. bustillo , j. w. simpson , g. t. mulhern and j. m. rothberg , _ dna transport by a micromachined brownian ratchet device _ , _ pnas _ * 96 * \ { 23 } ( 1999 ) 1316513169 .
d. heath , d. kinderlehrer , kowalczyk _ discrete and continuous ratchets : from coin toss to molecular motor _ , _ website : http//math.smsu.edu / journal _ _ discrete and continuous dynamical systems series b _ * 2 * \ { 2 } ( 2002 ) 153167 . | several authors @xcite have implied that the original inspiration for parrondo s games was a physical system called a `` flashing brownian ratchet @xcite '' the relationship seems to be intuitively clear but , surprisingly , has not yet been established with rigor .
the dynamics of a flashing brownian ratchet can be described using a partial differential equation called the fokker - planck equation @xcite , that describes the probability density , of finding a particle at a certain place and time , under the influence of diffusion and externally applied fields . in this paper
, we apply standard finite - difference methods of numerical analysis @xcite to the fokker - planck equation .
we derive a set of finite difference equations and show that they have the same form as parrondo s games .
this justifies the claim that parrondo s games are a discrete - time , discrete - space version of a flashing brownian ratchet .
parrondo s games , are in effect , a particular way of sampling a fokker - planck equation .
our difference equations are a natural and physically motivated generalisation of parrondo s games .
we refer to some well established theorems of numerical analysis to suggest conditions under which the solutions to the difference equations and partial differential equations would converge .
the diffusion operator , implicitly assumed in parrondo s original games , reduces to the schmidt formula for the integration of the diffusion equation .
there is actually an infinite continuum of possible diffusion operators .
the schmidt formula is at one extreme of the feasible range .
we suggest that an operator in the middle of the feasible range , with half - period binomial weightings , would be a better representation of the underlying physics .
physical brownian ratchets have been constructed and have worked @xcite .
it is hoped that the finite element method presented here will be useful in the simulation and design of flashing brownian ratchets . |
spiral structure is an astounding characteristic of disk galaxies . yet
, the nature of the spiral structure of our home galaxy is far from well understood .
the generally accepted model , originally discussed by @xcite , @xcite and @xcite , involves a density wave resulting from a global gravitational mode , triggered either by a internal instability or some driving element ( like a bar or an interacting external galaxy ) .
another proposed model describes the spiral structure as a self - propagating wave of enhanced star formation @xcite . in either case , the models show the importance of the gaseous disk in the global spiral structure phenomenon . the role of the gas in the spiral structure has been exploited repeatedly in order to trace the spiral arms , both in the milky way and in external galaxies .
those efforts have included @xcite , co clouds @xcite and dust @xcite as tracers .
since the spiral arms show an enhanced star formation rate , pop i objects can also be used as tracers . the model presented by @xcite , which featured a four - arm spiral traced by galactic regions ,
is frequently cited .
nevertheless , it is not unusual to observe differences in the spiral structure when different tracers are used .
one example is ngc 2997 , in which one clearly defined arm in optical light is absent in infrared light @xcite . in the milky way galaxy , @xcite found that the dust emission is better fit by a four - arm spiral , while a two - arm spiral fits the stellar emission ( * ? ? ?
* presents a review ) . as a result of the coupling between the gaseous disk and the interstellar magnetic field ,
the latter also is influenced by ( and influences ) the spiral pattern . since the magnetic fields are `` illuminated '' by the cosmic rays spinning along the field lines ,
the so generated synchrotron emission is a direct probe of the field . in external galaxies
, the total ( polarized + unpolarized ) synchrotron emission tends to follow the spiral pattern , although a displacement between the positions at which the optical and synchrotron emission peaks is not unusual .
in addition , analysis of the polarization direction of the synchrotron emission shows that the magnetic field is usually aligned with the spiral arms @xcite .
then , it can be argued that the large scale magnetic field plays a significant role in the formation of spiral structures in disk galaxies .
@xcite explored what effect a strong galactic magnetic field would have in the vertical structure of the spiral arms .
the main difference between their models and previous work was the inclusion of a thicker , higher pressure ism .
the necessity of such environment was pointed out by the fact that some components of the gaseous disk were observed to have scale heights of the order of kpc @xcite , much larger than those previously considered .
the support necessary for the weight of that gas is larger than the thermal pressure observed in the midplane , leading to the conclusion that non - thermal pressures must dominate the vertical hydrostatics of the galactic disk , and that the pressure scale height might be much larger that the density scale height @xcite .
when those pressures significantly increase the effective @xmath2 of the medium , the gas flowing into the spiral arms shows a combination of a shock and a hydraulic jump at the position of the spiral arms , which induced large vertical gas motions . such behavior appears to have been observed in ngc 5427 @xcite . in an earlier work ( *
* paper i from here on ) , we extended @xcite analysis to three dimensions and included a large fraction of the galactic disk .
it is the purpose of this work to further explore those results and aim more directly at the milky way structure .
the paper is organized as follows : section [ section_setup ] summarizes the results of paper i , presents the initialization of the simulations we present here , and describes the basic features of the flows at an age of @xmath3 ; section [ section_bfield ] describes the magnetic field structure , the synchrotron emission , and the field - density relationships ; section [ section_hydrostatics ] explores the elements of the support of the disk and the vertical hydrostatics near the solar circle ; section [ section_periodicity ] explores periodicity and normal frequencies in the simulation ; section [ section_circulation ] describes the large scale motions encountered in the simulation and evidence for circulation ; and section [ section_conclusions ] presents our conclusions .
details of the setup are provided in paper i. here we present only an overview , mentioning the differences from the cases presented there .
we performed three - dimensional mhd simulations using the code zeus @xcite to model the ism response to a spiral gravitational perturbation .
the gas starts in vertical and radial hydrostatic equilibrium , following the circular orbits defined by the background gravitational potential from model 2 in @xcite , the pressure gradient ( thermal plus magnetic ) , and the magnetic tension .
the magnetic field is azimuthal , with a strength defined by the relation : @xmath4 where @xmath5 is the magnetic pressure , @xmath6 and @xmath7 .
the equation of state for the initialization is isothermal with a temperature @xmath8 .
as shown in paper i , by defining the density profile in the midplane ( in our cases , an exponential with a scale length of @xmath9 and large @xmath0 . ] , with @xmath10 at @xmath11 ) , all the equilibrium variables are easily found .
again , the hydrostatics , the magnetic field geometry , and the density - magnetic pressure relation are only enforced in the initialization .
all of them are altered considerably by the mhd evolution .
the simulations were carried out in a cylindrical coordinate system .
the grid spans from @xmath12 to @xmath13 in the radial direction , 0 to @xmath14 in @xmath0 and 0 to @xmath15 in azimuth for the four - arm models ( 0 to @xmath16 in two - arm models ) .
both @xmath17 and @xmath0 boundaries are reflective , while periodic boundaries are set in the azimuth
. this initialization yields the distribution presented in figure [ initial ] . with slightly different parameters (
@xmath18 and @xmath19 ) , the vertical density at the solar circle would closely resemble the one described in @xcite .
we chose to use a lower magnetic pressure because we expected that the accumulation of the gas in the spiral arms would increase the magnetic field to values closer to those observed .
also , we expected that the gas dynamics could possibly generate a random component of the field ; this did not occur .
there are several possible reasons : low spatial resolution , numerical diffusivity , a low effective reynolds number , a near laminar flow , or a combination of all .
we chose to use a higher temperature to help offset the smaller magnetic pressure ( this was necessary because , without the extra pressure , the density drops very fast with @xmath0 , generating numerical problems ) .
it should be noted that our relatively high `` thermal '' pressure is a rough approximation to the pressures provided in fact by the random field , cosmic rays , subgrid turbulence , and the comparatively weak true thermal pressure . since this pseudo - thermal component is isothermal ( except at high density , see discussion below )
, it did not add the stiffness to the medium that the random field or other components would have .
as a result , we likely under - represent the jump - like character of the flow and the associated vertical and turbulent motions .
the rotation curve versus @xmath17 is nearly flat , increasing only slightly with @xmath0 . from equation [ mag_pres_eq ] , the alfvn velocity increases with distance from the midplane , to an asymptotic value of about @xmath20 .
after the simulation started , we changed the equation of state in order to facilitate the condensation of the gas into the spiral arms .
gas with @xmath21 behaves isothermally with @xmath22 ; at higher densities , the temperature of the gas is forced down , generating a thermally unstable regime , up to a density of @xmath23 , above which the gas behaves isothermally again , with @xmath24 , effectively forming a two - phase medium ( figure [ two - phase ] ) .
this change will have the biggest effect near the midplane , where thermal pressure dominates the magnetic pressure .
the choice of @xmath25 , rather than @xmath26 that one might expect from the heating - cooling balance , was made because this pseudo - thermal pressure represents other components that can not be radiated away , and because of our coarse resolution .
it is a modest attempt to sample the effects of condensation , either due to reduced thermal pressure or self - gravity .
in contrast to paper i , the calculation is performed in the reference frame of the spiral perturbation , which moves with an angular velocity of @xmath27 .
the perturbation locus is a logarithmic spiral with a pitch angle of @xmath28 . in radius ,
the potential perturbation varies in such a way that the implied density perturbation decays exponentially with a scale length of @xmath29 . in azimuth , the perturbation is sinusoidal , with a peak - to - valley amplitude corresponding to an arm / interarm mass ratio of 3.16 at @xmath30 . in the midplane ,
the perturbation force is a relatively constant fraction , @xmath31 , of the average radial force .
details of the radial and vertical shape of the perturbation are presented in @xcite .
the perturbation is turned on linearly during the first @xmath32 of the simulation .
also , to avoid splashing against the inner radial boundary , the perturbation forces are not applied in the @xmath33 , and smoothly increase to full strength in the next kpc .
therefore , the actual useful grid extends from 5 to 11 kpc . in summary , the differences between the models presented here and those of paper i are : the new equation of state , a higher magnetic pressure in the initialization , and the shift of the calculation to the pattern reference frame .
also , in order to avoid some of the numerical problems described in paper i , we set a density floor of @xmath34 , equal to the minimum density in the initialization .
we were consequently able to run the simulation to much greater ages .
the setup described should be susceptible to the parker instability @xcite .
nevertheless , we do not expect to see the interchange mode of the instability due to our low resolution , while the undular mode is expected to be quenched by both our lack of resolution and the flow we are modeling . in order to test this expectation , we restricted the computational domain to a two - dimensional grid along the solar circle , and set up an atmosphere like the one described above .
when we turned the arm perturbation off , we observed the growth of the undular mode of the parker instability .
but , when the perturbation was added , the magnetic field lines bent due to the vertical motion of the gas generated by the hydraulic jump , and not so much by the weight of the gas .
so , any vertical perturbation in the magnetic field will be carried along with the orbiting gas , and will be damped out when it encounters the hydraulic jump .
therefore , as long as the arm - encounter time is shorter than the instability growth time scale , the undular mode of the instability should not be present
. figures [ two_arm ] and [ four_arm ] show the structures of our two standard models ( two and four arms ) after 800 myr .
( unless otherwise noted , we report the state of the simulations at this time . ) in both figures , the grayscale in the upper panel shows half - disk column density ; the solid lines show the integrated pattern frame velocity field in the midplane .
the grayscale in the lower panels shows a density cut at a cylindrical surface with @xmath35 .
the arrows represent the velocity field in the pattern reference frame .
the length of the arrows is proportional to the total velocity in the @xmath35 surface ; but , since the vertical axis is greatly stretched ( in reality , the aspect ratio of the grid is @xmath36 ) the individual components of the velocity are also stretched so that the arrows point in the corresponding direction with respect to the density structures .
as these are models of trailing spirals , the gas rotates clockwise in the upper panels , and from left to right in the lower ones . in paper
i , we showed that , as the gas enters the arm , a combination of a shock and a hydraulic jump is formed .
the extra stiffness the magnetic field adds to the ism makes the gas jump over the obstacle the gaseous spiral arm represents .
the gas shoots up before the arm , forming an forward leaning shock , and falls back down after the arm , generating a secondary shock .
in isothermal cases with no magnetic field ( paper i ) , there is much less vertical motion , the forward shock is nearly vertical , and there is no secondary shock .
in the two - arm case , when the gas falls behind the arm , it bounces back up generating a high @xmath0 interarm structure that mimics another arm .
the high @xmath0 material then falls again before reaching the next arm .
this structure does not show up in the column density , but it is clear in both velocity and density in the lower panel of figure [ two_arm ] .
it is explored in the left panel of figure [ total_flux ] , where the contours show the zero - level of the integrated vertical mass flux ( @xmath37 ) , with the dashes pointing in the downflow direction .
the column density is shown in grayscale .
the downflow region is broader and happens at lower densities than the upflow . also , there is a second upflow region downstream of the arm , that mimics the flow structure around the arm .
a powerful bounce therefore appears to be a mechanism that might double the number of gaseous arms relative to the stellar arms , as is apparently observed in some galaxies .
generally speaking , in contrast with the model of @xcite , the gaseous arms sit downstream from the perturbation minimum , but follow a tighter spiral ( with a pitch angle of about @xmath38 ) .
we have not yet discovered the reasons behind either the pitch angle difference ( which appears in both 2d and 3d models ) or the phase difference relative to @xcite result .
the four - arm case ( figure [ four_arm ] ) has a similar structure to the two - arm one , but presents a big difference : when the gas falls behind the arms , it does not have enough space to complete a bounce and fall again before it encounters the next arm .
the downflow does lead to vertical compression ahead of the next arm , but not to a doubling of the vertical velocity pattern .
the leading shock is also more vertical , especially at higher @xmath0 .
the right panel in figure [ total_flux ] shows the vertical mass flux for the four - arm case . here , the upflow region is broader than in the two - arm case and ,
as noted above , there is no obvious interarm bounce structure .
( both two and four - arm models show higher numbers of features in narrow radial ranges , notably around @xmath39
. this may be due to a resonance between vertical oscillations and arm passage time , and is associated with broadening or even doubling of the arm structure . ) it is not clear how to separate the arm and the interarm regions .
we illustrate this in figure [ swollen ] .
as shown in the top panel ( a reproduction of the lower panel of figure [ four_arm ] ) , the high @xmath0 structure associated with a spiral arm is very wide , as traced by the density contours . in fact , the region over which the disk thins ( the contours reach low @xmath0 ) is very narrow . on the other hand , the column density ( second panel from the top )
also has clearly defined arms , but they cover a smaller region in azimuth than the high @xmath0 density structure .
the arms are even narrower when we look at the midplane density ( third panel from the top ) .
notice that , even if the gas flow does not bounce in the interarm , the downflow compresses the disk and generates an overdensity near the midplane before it encounters the next arm .
a frequently used measure for the thickness of the disk is the ratio of the column density to the midplane density , which is an estimate of the scale height .
examination of the top panel of figure [ swollen ] leads to the idea that the disk swells at the arms .
but , since the arm characteristics have different extents along azimuth , the estimated scale height ( bottom panel ) has a local minimum in the middle of the arm . in the leading and trailing end of the arm , the scale height increases , as expected .
this difference in the behavior of the column and midplane densities causes the scale height defined in this way to behave counter - intuitively . in figure [ four_arm ]
the arms seem to be discontinuous , with a break at about @xmath40 .
each of the sections follows an even tighter spiral ( with @xmath41 pitch angle ) than the perturbation ( @xmath42 ) and the overall locus of the arm ( @xmath38 ) .
the locus pitch angle has the advantage that the resulting arms are a good approximation to the milky way s spiral arms as traced by @xcite .
figure [ match_arms ] shows the column density of our four - arm model and the locus of the arms in the aforementioned work , scaled so that the distance from the sun to the galactic center is @xmath29 .
this agreement is extensively used in @xcite , where we generate synthetic observations using this model , for comparison with real milky way data .
figures [ two_arm ] and [ four_arm ] show that , as material in the midlpane approaches the arms , it moves radially outward , then shocks and follows a path nearly parallel to but soon outside of the arm , moving radially inward in the process . as we discuss below , the post shock inward flow at the midplane
is smaller than the outward preshock excursion , and there is a slight migration outward balanced by inflow at higher @xmath0 .
the magnetic field is carried along with this flow , and therefore has a radially outward component ahead of the arms and a radially inward component just outside of them . on figure [ v.b1 ] , these radial excursions of the field lines are plainly seen . at the locations of the arms , the field is roughly parallel to the arm locus . between the arms ,
the field pitch changes from somewhat less positive than the arm pitch , to circular , to negative pitch as it approaches the next arm . at higher
@xmath0 ( figure [ v.b2 ] ) , both the velocity pattern and field structure are somewhat less regular . as expected from the gas flow , the largest vertical component of the magnetic field is found just before and after the arms .
the vertical field can be as large as @xmath43 , but @xmath44 is a more typical value . the two - arm case ( not shown ) has a much smoother magnetic field .
it can also show negative pitch in the interarm region , but it does not change as abruptly as in the four - arm case . also , the vertical field is weaker , being prominent only on the leading edge of the gaseous arms .
owing to our missing the random field component , a synthetic synchrotron map can not represent the emission actually expected from a galaxy , but it could provide a reasonable view of the polarized fraction .
figure [ synch ] shows our calculated total emission , for our four - arm case , as viewed from outside the galaxy , with lines showing the directions of the apparent field ( perpendicular to the @xmath45 field polarization of the integrated emission ) , assuming no faraday rotation . at any given distance above the plane ,
the emissivity is given by : @xmath46 where @xmath47 is the component of the field parallel to the plane of the galaxy , @xmath48 is the spectral index of the electron cosmic ray distribution , @xmath49 is the spatial distribution of the electron cosmic rays , @xmath50 , and @xmath51 @xcite .
this emissivity is integrated along a line of sight perpendicular to the galactic plane .
the polarization is calculated by estimating the polarized emission parallel and perpendicular to @xmath47 , @xmath52 where @xmath53 is the degree of polarization of the emissivity . given the angle @xmath54 and the intensities @xmath55 , @xmath56 and @xmath57 at a certain @xmath0 , the intensities and parallel direction at @xmath58 are given by : @xmath59 } { ( i_\perp - i_\parallel)+ ( \varepsilon_\perp - \varepsilon_\parallel ) dz \cos[2(\beta-\alpha)]}\right ] \nonumber \\ & & \\ i'_{tot } & = & i_{tot } + \varepsilon_{tot } dz \\ i'_\parallel & = & i_\perp+\varepsilon_\perp dz -(i_\perp - i_\parallel)\cos^2(\alpha'-\alpha ) \nonumber \\ & & -(\varepsilon_\perp-\varepsilon_\parallel ) dz \cos^2(\alpha'-\beta ) \\
i'_\perp & = & i'_{tot}-i'_\parallel,\end{aligned}\ ] ] where @xmath60 is the direction parallel to the magnetic field at @xmath58 .
notice that @xmath61 is not the direction of polarization nor the direction of the magnetic field , but the direction of the inferred b - field , as traced by the synchrotron emission . in figure
[ synch ] we can see that the interarm turning of the b - field happens around the area of minimum emission , which is consistent with the picture of being a consequence of stretching the field . the higher intensity regions tend to be slightly upstream from the arm .
the degree of polarization is quite large ( about 70% ) .
this is expected , since our resolution does not allow us to model the random component of the field , that might depolarize the emission .
only a shift in the direction of the b - field with @xmath0 reduces the polarization from the @xmath62 of the intrinsic emissivity . for our two - arm case ( not shown ) ,
the polarization changes are not as abrupt , the arms are more continuous ( no blobs with associated peaks of synchrotron emission ) , and the difference in the positions of the peaking of emission and column density is smaller .
figure [ synch ] is to be compared with the maps from @xcite . in that work ,
an atlas of synchrotron emission in external galaxies is presented .
they compared the polarized and total emissions , as a way of comparing the total ( random + regular ) and regular b - fields ( again , we do not have enough resolution to say much about the random field ) .
they consistently found islands of synchrotron emission and displacements between the synchrotron and gaseous arms not so different from the ones found here .
we did not find magnetic arms without a density counterpart , which have been occasionally found in polarized emission observations and some simulations ( * ? ? ?
* for example ) .
galaxies should have a lot of random field in the spiral arms , generated by both the turbulence of the jump and stellar feedback , but the post arm flow should stretch the field to greater regularity in the interarm region .
to the extent that this is true , we would expect the mean convected field to follow the pattern just described .
some observations @xcite suggest that in some galaxies there are magnetic arms , regions of enhanced polarized synchrotron emission not associated with the gaseous arms .
our finding that in two - arm spirals , with their longer interarm transit , the flow tends to bounce and generate an arm - like structure in the interarm region could also be related to this phenomenon .
a detailed exploration of these points is outside the scope of the present work .
the relationship between magnetic field strength and density in the interstellar medium is an interesting diagnostic of the behavior of the flow . to the extent that the various components of the ism are all just the current states of material that samples all states , so that there is no intrinsic relationship between density and flux
, the relationship should distinguish the various dynamical possibilities .
if the field were small enough to be dynamically insignificant , compressions on average should be three dimensional , which with flux freezing would yield @xmath63 .
if the field were so strong that it dominates the pressure , compressions would be possible only along the field , and there would be no dependence of field strength on density , except in dynamically active environments .
measurements of the field strength in the midplane , at average and lower densities seem to find no dependence of @xmath64 with @xmath65 @xcite . at densities high enough that local dynamics or self gravity is important , higher field strengths ( and higher total pressures ) are found @xcite . at the lower densities ,
some measurements provide only upper limits to local region fields , but the strong fluctuations in pulsar dispersion measures and the overall synchrotron emissivity seem to imply that the rms field is several microgauss .
it would appear that most of the volume , even at low density , must have fields of this order ( see discussion in valle , 1997 ) .
our simulation can be explored for its @xmath66 relationship , though some care is required in interpretation .
two factors must be kept in mind .
first , we initialize our run with a very strong dependence of @xmath64 on @xmath65 , as given by equation [ mag_pres_eq ] .
it is almost certain that far from the plane of the galaxy , not only is the density low , but so is the magnetic field .
our initialization is a specific rule for this relationship , and determines the static vertical structure .
we observe , however , that even in our relatively laminar flow , material makes large vertical excursions in its response to the arm disturbance ( see section [ section_circulation ] ) .
thus , the material found at any given height will have a wide spread in initial heights , and therefore , a wide spread in initial @xmath66 .
this is more true further from the plane , but should not be discounted close to the plane as well .
the second point is that because we do not have a significant random field , it may be easier for material to move along the field than in the actual galaxy .
this is not necessarily true if , for example , most of that motion occurs in interarm regions where the field is less random ( as discussed above ) .
it is also possible that in the actual galaxy , a considerable amount of motion along fields is driven at small scales and high densities by stellar feedback in the arms , and then frozen in as the material and field are stretched out between the arms , the field returning to its pressure defined equilibrium , with whatever distribution of density is left from the small scale dynamics . in short , the results we are
about to present are suggestive , but can not be assumed to represent reality very precisely .
owing to our strong @xmath66 in the initialization , we restrict our presentation to material found between 7 and @xmath67 , and present three ranges in @xmath0 , close to the plane , intermediate , and high , the boundaries between them being at 50 and @xmath68 .
we also separate the material into two equal halves ( in space ) , one containing the arms , the other not .
the results are shown in the six panels of figure [ mag_dens_arms ] .
the contours are of volume containing @xmath66 .
there are also several lines on these diagrams , one showing the @xmath66 of the initialization , one showing the locus of equal magnetic and thermal pressures , and one ( shown only on the midplane arm plot ) showing a line of constant total pressure of @xmath69 .
there are several things to notice .
first , at no height does the initial trend of @xmath66 represent the trend of the data , though in every diagram , significant amounts of material are found along parts of the initialization .
next , there are three characteristic features found in the plots , sections where the distribution has a major horizontal feature , sections where there is a major feature or slope somewhat steeper than the initialization , but lower than the slope ( one ) expected from purely transverse compression , and in the midplane , a feature at the highest densities which has a concave upward boundary .
the horizontal segments , most noticeable at the intermediate heights , are exclusively found in regions in which the magnetic pressure is higher to significantly higher than the thermal pressure .
our flow thus seems to reproduce the expectation that in such regions the dominance of magnetic pressure is accompanied by flows parallel to it .
the vertical width of the horizontal segment should then be associated with the spread in pressures found between 50 and @xmath68 in the central pair of panels .
the width is roughly a factor of 1.4 in @xmath64 and therefore 2 in pressure , more or less consistent with the pressure variation with height presented in section [ section_hydrostatics ] .
the fact that this behavior is not well represented in the midplane is likely due to our initialization with dominant thermal pressure at the midplane densities .
notice that in our simulation , the densities within @xmath70 of the midplane are never as low as those expected for intercloud regions in the milky way .
the inclined segments have @xmath71 or so , roughly as expected from isotropic compression , a little steeper than the initialized relationship . instead of showing isotropic compression
, it could be a mixture of the initialization and the @xmath72 expected from purely transverse compression .
we do not therefore give definitive reasons for this behavior , but note that it is always found in material whose magnetic pressure is less than or comparable to its thermal pressure .
it is in the regime for which isotropic compression is plausible .
the third feature , the concave upward behavior at the highest densities close to the plane , appears to be a consequence of our thermal equation of state .
this behavior occurs at those densities for which the temperature is rapidly decreasing with increasing @xmath65 , requiring a higher value of @xmath64 to have the same total pressure .
notice the similarity of the locus to the constant pressure line superimposed .
various authors ( * ? ? ?
* and references therein ) have attempted to reconcile the apparent midplane pressure of the galaxy with the weight of the material above it . in this section ,
we test two aspects of this procedure , whether such equality is expected locally ( for example , in the immediate neighborhood of the sun , there appears to be a scarcity of material relative to the average and one might wonder whether this should be reflected in a lower than average midplane pressure ) , and the degree to which dynamics contributes to the overall balance . in figure [ pressure ]
we compare the midplane thermal plus magnetic pressure with the integral of the overlying weight per unit area versus phase along a line of constant galactocentric radius of 8 kpc .
the half - disk column density is also provided for comparison .
the mean value of the midplane pressure ( @xmath73 ) is in good agreement with the mean weight ( @xmath74 ) .
but the behavior with phase shows much more interesting structure .
the midplane pressure does have significant variations , as does the weight , and they are correlated to some degree .
a perfect match is not expected because material does not evolve on constant radius , because these results are for one specific time , during which things might be slightly out of equilibrium , and because the pressure calculated does not include dynamical contributions .
still , it is possible to get a sense of the dynamical behavior from this plot .
for @xmath75 to 1 , the column density is nearly constant , so the weight is a measure of the vertical extent . from left to right , the weight falls , as does the material , raising the midplane density and pressure .
the trend reverses about @xmath76 to upflow , increasing weight and decreasing pressure .
this is a bounce .
then , there is a sudden increase in pressure and weight , a shock , that raises the pressure above the weight , though only slightly .
this is the hydraulic jump .
it sustains the upflow , roughly stabilizing the increased height .
thereafter , the flow accelerates parallel to the plane , decreasing the pressure , weight , and column density .
because the effective @xmath2 is grater than 1 , the pressure drops faster than the weight , leading to downflow , a downstream pressure increase and weight minimum , ready to bounce again .
we have explored the vertical hydrostatics more precisely , averaging over phase to see how the contributions to pressure versus height are arranged .
we can not explore the importance of magnetic tension as @xcite did , since our model has restrictions that limit the development of vertical magnetic field , namely , lack of spatial resolution , no cosmic rays pushing the magnetic lines , nor an effective dynamo .
nevertheless , we found that the large scale motions of the gas provide a significant fraction of the support .
consider the vertical component of the momentum conservation equation : @xmath77 = -\rho g_z - \frac{\partial}{\partial z}\left(p_{th}+p_m\right),\ ] ] where @xmath78 is the component of the velocity in the vertical direction , @xmath79 is the vertical gravitational acceleration , @xmath80 is the gas density , @xmath81 is the thermal pressure , and @xmath82 is the magnetic pressure . here , we used the approximation proposed in @xcite , in which the magnetic pressure is diminished by the magnetic tension : @xmath83 . in our model ,
the vertical field is not very large , and the magnetic tension is only a small fraction of the vertical support . in the case of @xmath84 , @xmath85 in figure [ pres_terms ] ,
the left panel shows the four different integrated terms in the right side of equation [ hydro_2 ] , and the right panel compares the left and right hand sides of the equation . in that figure
, @xmath86 is the value of the thermal plus effective magnetic pressure at the upper boundary at @xmath87 .
all the plotted quantities are actually the mean values along the solar circle , so that the effect of the spiral arms is averaged out to give a global picture . in order to calculate the weight of the gas , the total gravitational potential ( background + perturbation )
was used , but it makes little difference , since the actual value of the perturbation is small compared with the background potential .
figure [ pres_terms ] shows that the spatially averaged disk atmosphere is actually very close to hydrostatic at the @xmath88 time shown .
although most of the pressure is balanced by the weight of the gas , the convective terms have a significant influence , specially at intermediate @xmath0 . as expected , the first two convective terms work into supporting the weight of the gas , generating kinematic pressure . but
the fourth term , the vertical kinematic pressure @xmath89 , has the same sign as the gas weight .
this is because its average is biased toward the highest vertical velocity regions .
such regions are the downflow behind the arms , where @xmath78 is negative and becoming more negative with increasing @xmath0 . as hinted by figure
[ pressure ] , the midplane disk pressure has to decelerate the downflow in addition to supporting the weight of the gas . at different times ,
the lines showing the weight of the gas ( dashed line in the right panel of figure [ pres_terms ] ) and pressure ( solid line ) oscillate only slightly around each other , suggesting that the vertical hydrostatics described above is not atypical .
the simulations do not seem to be in a steady state , but rather in a steady cyclic behavior , in which the `` wave - like '' structure that is formed above the gaseous arms breaks , only to be regenerated at the back of the arm and rise again .
the periodicity of the simulation can be assessed by plotting the vertical mass flow , at a given height , integrated over the @xmath90 plane , as a function of time .
such a plot for the four - arm case is presented in figure [ mass_flux ] .
the chosen height is @xmath91 , which is right across the `` breaking wave '' structure ( see figure [ four_arm ] ) .
the left panel of figure [ mass_flux_fourier ] shows the fourier transform of this vertical flux for @xmath92 .
it shows a strong peak at about @xmath93 . that peak grows smaller
if we restrict the integration to only a range of @xmath14 around @xmath35 ( @xmath14 is the typical width of the gas radial motions , see section [ section_circulation ] ) .
the right panel of figure [ mass_flux_fourier ] shows that case .
notice that a second strong peak at @xmath94 appears .
both peaks are present when we move the integration range to @xmath95 and @xmath96 , although they both move to slightly smaller periods and the large period peak becomes weaker than the short period one as we go to smaller radii .
we performed a linear perturbation analysis of the vertical motions of the disk in order to understand the origin of these periods .
the procedure is very similar to the one followed in the appendix of @xcite .
consider the equation of motion in the vertical direction for an plane - parallel isothermal atmosphere : @xmath97 where @xmath98 is the velocity , @xmath80 is the mass density , @xmath99 is the total pressure ( thermal plus magnetic ) , and @xmath100 is the gravitational acceleration .
consider perturbations to the ( hydrostatic ) equilibrium distribution , @xmath101 substituting these into equation [ eq_motion ] , and eliminating the hydrostatic equilibrium condition , we get : @xmath102 where @xmath103 . using mass conservation , @xmath104 and assuming that @xmath105 , we can eliminate @xmath106 to get , @xmath107 + v \left [ \frac{\rho_0''}{\rho_0 } - \left(\frac{g + ( c^2)'}{c^2}\right)^2 + \frac{\omega^2}{c^2 } \right ] = 0,\ ] ] where @xmath108 .
( notice that this equation reduces to a sound equation for very large frequencies . )
equation [ v_pert_eq ] can be solved numerically for the initial density and pressure conditions of the simulation , for a given frequency @xmath109 .
we explored a range of values for @xmath109 and looked for solutions to equation [ v_pert_eq ] that are consistent with the zero vertical velocity boundary conditions we imposed in the simulation .
the lowest frequency consistent with the boundary conditions corresponds to a period of about @xmath32 for the vertical gravity and density structure at @xmath35 . as with the analysis from figure [ mass_flux_fourier ] , that period decreases as we move inward ( @xmath110 at @xmath95 ) , and increases outward ( @xmath111 at @xmath112 ) .
we did not find a period corresponding to the other @xmath113 peak , which could involve radial motions .
in this section , we look for global trends in the gas motions to explore circulation and radial motions of the material . in order to do so , we averaged all the velocities and densities of the simulation for the period of @xmath114 . then , we integrated the resulting mean velocity field in order to get the trajectory of an imaginary test particle .
the result for the four - arm case is presented in figure [ paths ] . in this , the grayscale is the time averaged density at @xmath115 .
the lines show the trajectory in @xmath17 and @xmath0 that a test particle follows in the mean velocity field in going from @xmath115 to @xmath15 ( one arm encounter time , which differs with radius ) , ending at the position with the larger dot .
the first thing to notice in this diagram is that , as the gas goes between arm and interarm regions , it moves a typical distance of about @xmath14 in @xmath17 , and a varying amount in @xmath0 ; sometimes , the vertical displacements can be as large as @xmath116 .
also , the trajectories have a slanted appearance .
this is consistent with what has been mentioned before : at the spiral shock , the gas shoots up and then moves radially inward along the arm . after leaving the arm ,
the gas falls down and moves out in radius as it approaches the next arm .
the difference between the roughly looping orbits seen at low @xmath0 and the prostrate `` s '' shapes at higher @xmath0 is that material at low @xmath0 is moving outward at minimum height , whereas at high @xmath0 it is moving inward .
reference to figure [ four_arm ] shows that , at low @xmath0 , minimum height occurs at the interarm bounce , where material is moving outward , while at high @xmath0 , minimum height occurs downstream of the leading edge of the arm , where material has already begun to move in . in figure [ arrows ] , we plotted an arrow from the initial to the final positions of the test particle .
the interpretation of this diagram is not straightforward .
each parcel is followed through one fourth of a rotation , relative to the pattern .
it typically ends up at both a different @xmath0 and @xmath17 , and at a different @xmath17 it is in a different phase with respect to the arm , as evidenced by the shift relative to the density distribution shown .
the pattern is also incomplete , in that it shows no obvious source of replenishment for material above @xmath117 outward of @xmath118 .
what is clear is that there is a general circulation pattern counterclockwise about each arm structure , in which material above @xmath68 has large radial and vertical net excursions .
higher @xmath0 material tends to move inward , and lower @xmath0 outward , though at lower velocity and higher density .
in the upper regions of the inner edge of each arm , there is some indication of inward motion that transits from one arm to the next , but we are reluctant to conclude , from modeling by this eulerian code , that there is thorough mixing of material across the disk . the fact that the net circulation pattern has several cells , one associated with each arm , shows that it is not a simple consequence of our closed inner and outer boundaries , which might have led to a single cell of circulation .
this figure is ambiguous about what happens to material very close to the plane , leading us to study the average radial velocity of material as a function of @xmath0 .
we calculated the mean radial velocity of the gas at different heights by averaging the radial velocity at a given @xmath0 , weighted by the density at that point : @xmath119 in figure [ radial_flow ] , we present this mean radial velocity for the time averaged four - arm case .
the velocities are inward and significant above @xmath120 , reaching @xmath121 at about @xmath122 . near the plane they are outward and low , only about @xmath123 is reached .
this is consistent with the observational limits ( * ? ? ?
* and references therein ) and inferences from the previous figure . from this ,
we conclude , as anticipated , that there is net outward flow close to the plane to compensate the net inward flow at higher @xmath0 .
but we are still uncertain whether this flow occurs in essentially closed cells , separately mixing material within each arm , as suggested by figure [ arrows ] for most of the material , or whether there is sufficient transit between arms for substantial global mixing .
our tentative conclusion is that the velocities shown in figure [ radial_flow ] are dominated by closed cell circulation , and therefore represent a strong limit on the global mixing rate .
we performed mhd simulations of the gaseous galactic disk having spiral gravitational perturbations with two and four arms .
as found in @xcite and paper i , the gas motions involve vertical bouncing , plus a combination of a shock and a hydraulic jump at the gaseous arms .
the comparative timescales for bouncing and interarm passage , and the realtive phases lead to complications in the details of the arm structure , as predicted by @xcite and @xcite .
this bounce / shock / jump structure leans upstream above the plane , more noticeably in the two - arm case , where the bounce point occurs within the arm at low @xmath0 . in the four - arm case ,
the low @xmath0 bounce point occurs at the interarm density peak ( at @xmath124 ) and the upward motion from the bounce leads the shock / jump structure . in both cases , the gas moves up and over the midplane arm , inward in radius .
the higher @xmath0 material then falls down behind the arm , where it is again moving radially outward . in the two - arm case
, the gas bounces back up and , at high @xmath0 , generates structures similar to those at the arm , as if the gas were trying to develop a four - arm structure .
in the four - arm case , if the sun s position is properly chosen , the arms seem to follow the arms traced by @xcite . in both cases , the gaseous arms follow a spiral tighter than the imposed perturbation . since the net radial inflow happens above the arms , and the gas falls in the interarm , when it is moving out before encountering the next arm , the disk averaged radial flow changes sign with height above the disk , from positive near the midplane , to negative higher up .
this cycle appears to happen in cells associated with the arms , and might not represent a global mixing phenomenon . within the spiral arms ,
the magnetic field adopts a pitch angle similar to that of the arms , but it develops a negative pitch in the interarms .
the vertical magnetic field is only important at the position of the largest vertical flows .
our model does not include cosmic rays , supernovae or other energetic events , and therefore , it does not develop much vertical field , and the total field strength falls faster with @xmath0 than observations suggest . those shortcomings , on top of our low resolution , did not allow us to model the random component of the field , which causes us to overestimate the synchrotron polarization fraction .
an examination of the relationship between @xmath64 and @xmath65 found that , when the magnetic pressure strongly exceeded the thermal , @xmath64 was constant , independent of @xmath65 .
conversely , when thermal pressure is significant , @xmath64 tends to increase with @xmath65 .
the disk atmosphere is close to hydrostatic equilibrium , when the dynamical terms are taken into account .
it is noticeable that the downward vertical ram pressure of the gas , when averaged in azimuth , plays an important role in the hydrostatics , incrementing the effective weight of the gas .
magnetic tension plays only a small part in the vertical support , an effect of the model s weak vertical fields , that may not represent the true situation .
the periodicity of the structures found in this model can be estimated by fourier analyzing the vertical mass flux .
we found two outstanding periods , @xmath125 and @xmath126 , the smaller of which can be explained in terms of the lowest normal mode of vertical oscillation .
modelers have the huge advantage of knowing exactly where material being studied is , and its velocity .
observers do not have that luxury .
we will make that connection for our model in @xcite , in which we generate synthetic observations , and then explore them from the advantageous point of knowing the details of the underlying distributions .
we thank r. benjamin , e. zweibel , m. martos and b. pichardo and the anonymous referee for useful comments and suggestions , to the nasa astrophysics theory program for financial support under the grant nag 5 - 12128 , and to mxico
s consejo nacional de ciencia y tecnologa for support to g. c. g. | as discussed in @xcite , the extra stiffness that the magnetic field adds to the ism changes the way it reacts to the presence of a spiral perturbation . at intermediate to high @xmath0 , the gas shoots up before the arm , flows over , and falls behind it , as it approaches the next arm .
this generates a multicell circulation pattern , within each of which the net radial mass flux is positive near the midplane and negative at higher @xmath0 .
the flow distorts the magnetic field lines . in the arm region ,
the gas flows nearly parallel to the arm , and therefore , the magnetic field adopts a similar pitch angle . between the arms , the gas flows out in radius , generating a negative pitch angle in the magnetic field . the intensity and direction of the field yield synthetic synchrotron maps that reproduce some features of the synchrotron maps of external galaxies , like the islands of emission and the displacement between the gaseous and synchrotron arms . when comparing the magnitude of the field with the local gas density , two distinctive relations appear , depending on whether the magnetic pressure is dominant .
above the plane , the density structure develops a shape resembling a breaking wave .
this structure collapses and rises again with a period of about @xmath1 , similar to that of a vertical oscillation mode .
the falling gas plays an important part in the overall hydrostatics , since its deceleration compresses the low @xmath0 gas , raising the average midplane pressure in the interarm region above that provided by the weight of the material above . |
the determination of the nuclear mass composition of uhecr ( ultra high energy cosmic rays ) is fundamental to unveil the origin of the most energetic particles known in nature .
uhecr are detected by means of the extensive air showers created in the earth s atmosphere , which are composed by a cascade of hadrons and electromagnetic ( em ) particles .
the depth at which the em shower reaches its maximum , @xmath0 , strongly correlates with the depth where the primary firstly interacted .
the @xmath0-distribution carries information about the primary particle and the physical processes in the cascade .
the hadronic cascade is composed mostly by pions , of which the charged pions might decay in flight into a muon and a neutrino . due to their long lifetime and low cross section , muons can leave the core of the hadronic cascade traveling kilometers away and be detected .
electrons spread out in time and space because of coulomb scattering , whereas muons practically travel following straight lines .
the time structure of the shower disc encodes the history of the shower , allowing us to recover information about the longitudinal evolution by analyzing the time distribution of the particles arriving at ground level .
the pierre auger observatory , located on the high plateau of the pampa amarilla , is the largest cosmic ray observatory ever built .
its hybrid design allows to collect the shower particles by a surface detector ( sd ) and to observe the longitudinal development of the em profile by collecting the uv light with a fluorescence detector ( fd ) .
the sd spans 1600 cherenkov detectors in a 1.5 km triangular grid over 3000 km@xmath1 , whereas the fd is composed by 24 telescopes distributed over 4 sites overlooking the array .
the baseline design is being complemented with enhancements both on the sd and fd , and new detection techniques @xcite .
in this paper we present the latest @xmath2 and rms-@xmath0 results as a function of energy , and the @xmath0-distributions for the highest and lowest energy bins .
we also present the evolution with energy of sd observables extracted from the time distributions of the signal which are sensitive to the longitudinal evolution of the shower .
we compare our results with the predictions of hadronic interaction models @xcite .
we have used data taken between december 2004 and september 2010 , following the analysis reported in @xcite and updated in @xcite .
we have considered those showers reconstructed by the fd that have at least a signal in one of the sd stations .
the longitudinal profile was fitted with a gaisser - hillas function , and a series of cuts depending on the atmospheric optical conditions and on the quality of the reconstructed profile were applied @xcite . in order to avoid that the data sample is biased regarding to the nuclear mass , a number of cuts in the geometry were imposed .
the resolution in @xmath0 is around 20 @xmath3 .
uncertainties in the atmospheric conditions , calibration , event selection and reconstruction result in a systematic uncertainty of @xmath4 @xmath3 .
the average values as a function of the energy are displayed in the left panel of fig .
[ em_assym ] ( third plot from top ) , alongside with the predictions for different hadronic interaction models . + + , for the lower and highest energy bins .
the mixed composition corresponds to 50% p and fe.,width=264 ] , for the lower and highest energy bins .
the mixed composition corresponds to 50% p and fe.,width=264 ] the elongation rate @xmath5 is better described using two slopes , with @xmath6=7.4/9 , compared to 54/11 if we used a single power law .
the change of the elongation rate could be interpreted as a change on the nuclear mass composition , provided that there is no change on fundamental properties of the hadronic physics .
the comparison of @xmath7 with the different predictions of the hadronic interaction models could be interpreted as a change of composition towards heavier elements at the highest energies .
the left panel of fig .
[ em_assym ] also displays the rms-@xmath0 , which has been corrected by the detector resolution .
it rapidly decreases with energy around the same energy of the change on the elongation rate .
the distributions of @xmath0 for all energy bins have been published in @xcite .
[ xmaxmodels ] plots the @xmath8-distributions for the highest and lowest energy bins .
the predictions of the different hadronic models show a nearly universal shape . at low energies ,
the shape of the data distribution is compatible with a very light or mixed composition , whereas at high energies a heavier composition is favored .
the high @xmath0 tail , characteristic of very light components , is suppressed with respect to the lowest energies .
in each sd event , the cherenkov detectors record the signal as a function of time .
the risetime , @xmath9 , defined as the time elapsed between the 10% and 50% of the integrated signal , depends on the distance to the shower maximum @xcite , the zenith angle @xmath10 and the distance to the core @xmath11 . in inclined showers ,
the average risetime depends on an angle defined on the perpendicular plane @xmath12 as @xmath13 , due to the different effective distances of the tanks to the shower maximum .
the evolution of @xmath14 with zenith angle reaches a maximum at @xmath15 which is different for different primary masses @xcite . in the right central panel of fig .
[ em_assym ] an example of @xmath14 as a function of @xmath16 is shown for the energy bin @xmath17 .
events with energy above @xmath18 ev and @xmath19 were selected @xcite .
results of @xmath15 as a function of energy are displayed in the left panel of fig .
[ em_assym ] .
the systematic uncertainty amounts to @xmath20 10% of the proton - iron separation predicted by models .
the number of muons predicted by the hadronic interaction models differs from data @xcite . a preliminary study including this effect indicates a possible change of about @xmath21% of the proton - iron difference .
an approximated relation between the arrival time delay of a muon with respect to the shower front plane and the distance of the muon production point to ground can be used to transform the arrival time distributions of muons into production depth distributions @xcite . in @xcite ,
this technique was applied to auger data in an angular window between @xmath22 and @xmath23 for stations with @xmath24 m , as a result of a trade off between em contamination rejection and the intrinsic resolution of the method . in fig .
[ em_assym ] , right top panel , an example for a real event at @xmath25 eev is shown . similarly to the em component , the depth at which the muon production rate reaches a maximum , @xmath26 is a good indicator of the primary mass composition . after a series of cuts @xcite ,
the measured values of @xmath27 are presented in the upper panel of fig .
[ em_assym ] . the systematic uncertainty due to reconstruction bias , core position , rejection of the em component and quality cuts amounts to 11 @xmath3 , corresponding to 14% of the proton - iron separation .
note that the discrepancy on the total number of muons observed between models and data does not affect this result , but only the normalization of the muon production depth distribution .
provided that the present models give a fair description of the physical processes and their systematics , the average primary composition of uhecr could be inferred from the left panel of fig .
[ em_assym ] , showing a change towards heavier mass composition at high energies .
the evolution of @xmath7 , @xmath15 , and @xmath27 is similar in the overlapping energy regions , despite the fact of having completely independent systematics .
rms-@xmath0 evolution can be accommodated with a variety of compositions since it is influenced by the shower - to - shower fluctuations and the relative differences on @xmath7 .
the analysis of the shape of the @xmath0-distributions also points to the direction of heavier composition as we increase the energy . on the other hand ,
it is not guaranteed that current hadronic interaction models describe well particle physics at these energies @xcite , and a different interpretation would therefore apply .
9 h. klages for the pierre auger collaboration , these proceedings n.n .
kalmykov , s.s .
ostapchenko , and a.i .
pavlov , nucl .
suppl . * 52b * ( 1997 ) 17 - 28 ; e.j .
ahn _ et al_. , phys .
* d80 * ( 2009 ) 094003 ; t. pierog and k. werner , phys . rev .
* 101 * ( 2008 ) 171101 m. t. dova , m. e. mancenido , a. g. mariazzi , h. wahlberg , f. arqueros , d. garcia - pinto , astropart .
* 31 * ( 2009 ) 312 - 319 .
the pierre auger collaboration , nucl .
instrum .
* a613 * ( 2010 ) 29 - 39 . | the fluorescence detector of the pierre auger observatory measures the atmospheric depth , @xmath0 , where the longitudinal profile of the high energy air showers reaches its maximum .
this is sensitive to the nuclear mass composition of the cosmic rays .
due to its hybrid design , the pierre auger observatory also provides independent experimental observables obtained from the surface detector for the study of the nuclear mass composition .
we present @xmath0-distributions and an update of the average and rms values in different energy bins and compare them to the predictions for different nuclear masses of the primary particles and hadronic interaction models .
we also present the results of the composition - sensitive parameters derived from the ground level component . |
our main source of information about two photon interactions is the classic weizscker - williams process @xmath0 , where for exclusive channels @xmath1 is typically two to six mesons or a baryon - antibaryon final state .
the bulk of the data is obtained without tagging .
then , thanks to the behaviour of the photon propagator , the photons are almost on - shell . by factoring off the known @xmath2 vertices , we learn about real @xmath3 reactions
see for example [ 1 ] . with tagging one can , of course , also investigate important @xmath4 collisions .
i will not review these , but rather refer to the later talk of bernard pire [ 2 ] .
we can also study the time reversed process , as in e835 at fermilab , by looking at @xmath5 [ 3 ] . now consider the @xmath6 cross - section as a function of the @xmath3 c.m .
energy @xmath7 .
a typical example is shown in fig . 1 .
there one sees the cross - section rise from threshold , then have structure and subsequently decline .
this cross - section naturally divides into three kinematic regions which correspond to three different dynamical regimes . in each case
the photon couples to the electric charge of a point - like object , but what it sees as point - like changes with energy . at low energy , close to threshold
, the photon has long wavelength and sees the whole of the final state hadron and couples to its electric charge .
this teaches us about _
hadron dynamics_. as the energy increases and the wavelength of the photon shortens , it sees the charged components of the hadron , the constituent quarks . coupling to them
, it causes them to resonate , so we can learn about _ resonance dynamics_.
lastly , as the energy rises still further , the photon sees charged point - like objects inside the constituent quarks , the current quarks , and we can learn about _
quark dynamics_. the extent of the three kinematic regions depends on the final state . for pions ,
the three regions are well - separated , for kaons the near threshold region is foreshortened and the resonance regime more structured . in all three cases , the dynamics
is governed by qcd .
region 3 is the perturbative non - perturbative interface , while the lower two depend wholly on the strong physics aspects of qcd . for charmonium and charmed particle production ,
regions 2 and 3 merge with each other , both being amenable to perturbative treatment .
let us discuss region 3 , the higher energy regime first , as that is the easier to explain .
above a few gev , when two photons collide in the centre - of - mass frame , they deposit all of their energy in a region the size of a fraction of a fermi .
this creates a @xmath8 pair moving back - to - back .
these radiate soft gluons , until they reach separations of the order of a fermi , when the quarks and gluons fragment .
this produces two back - to - back jets of hadrons that dominate the inclusive high energy cross - section . to produce an exclusive final state like two mesons , the initial @xmath8 pair must radiate at least one hard gluon , which in turn creates another @xmath8 pair , moving back - to - back parallel to the initial quarks , so that these can get together to form a meson
it was brodsky and lepage [ 7 ] , who long ago , recognised that such processes can be divided into a short - distance part governing the emission of the hard gluon and a long distance component determined by the wavefunction of the hadron final state .
thus they predicted that the differential cross - section for @xmath9 should be given by @xmath10 where the wavefunction of the pion is related to the pion s electromagnetic formfactor @xmath11 .
this prediction ( fig .
2 ) is in reasonable agreement with data from mark ii [ 1,8 ] , tpc / two - gamma [ 1,9 ] , cleo [ 6 ] , et al . to form baryons in the final state
, a further hard gluon has to be radiated , to produce yet another back - to - back @xmath12 and @xmath13 .
it was farrar , maina and neri [ 11 ] , who first extended this discussion to @xmath14 production , by building in the proton s three quark wavefunction of chernyak and zhitnitsky [ 12 ] .
as seen in fig . 2 , this makes a prediction well below the data .
though baryons are built of three valence quarks , they spend part of their time in a quark - diquark configuration .
perhaps baryons are produced by hard gluons creating just one diquark - antidiquark , instead of two @xmath8 pairs .
this gives the predictions [ 13 ] shown in fig . 2 , for both the energy and scattering angle dependence , in far better agreement with data .
unfortunately , this detailed modelling is at variance with more recent cleo data on @xmath15 production [ 14 ] .
if we now look at the energy region of @xmath16 gev , we see distinct resonance structures , for instance in @xmath17 [ 15 ] in fig .
three well - known spin two resonances , the @xmath18 , @xmath19 and @xmath20 appear .
tensor mesons always appear strongly in vector - vector interactions . in the @xmath21 channel ,
the @xmath18 and @xmath19 overlap , so the data display an interference effect .
we know these mesons belong to an almost ideally mixed quark model multiplet , but nevertheless we can use the two photon data to test this . the photons couple to the electric charges of the quarks .
thus the two photon amplitude , fig .
4 , measures their mean charge squared and the cross - section ( and radiative width ) gives the square of this .
thus @xmath22 where @xmath23 is the probability that the quarks annihilate .
it is natural to assume for such a multiplet that the @xmath23 are the same for each member
. then the quark model simply predicts @xmath24 which is sensitive to small deviations from ideal mixing .
experiment confirms the closeness to this pattern with @xmath25 .
so much for relative rates , but what about absolute predictions for the radiative widths .
for these one needs a modelling of the dynamics .
the simplest to imagine is by analogy with positronium decay .
we begin with a fermion - antifermion bound state with definite quantum numbers @xmath26 .
this decays by one fermion exchange , as shown in fig .
this is readily calculated as shown by godfrey and isgur [ 16 ] , ackleh and barnes [ 17 ] , klempt et al . [
18 ] , resag and mnz [ 19 ] , schuler et al .
[ 20 ] , and many others .
the result simplifies for heavy fermion systems , where one can safely take the non - relativistic limit .
then the bound states formed by the interquark potential have radial wavefunctions @xmath27 that can be straightforwardly calculated using the schrdinger equation .
this gives , for instance , for spin singlet states this involves the @xmath29th derivative of the wavefunction at the origin , which is computable for charmonium , assuming the form of the interquark potential , coulombic plus linear confinement , for example .
note that the mass of the fermion enters , but not the mass of the bound state . while for heavy flavour systems it may seem eminently sensible to assume the _ effective _ mass of the quark is half that of the bound state , for light quark mesons , this is not so obvious . the relativistic analogue of eq .
( 4 ) is given in [ 17 ] for instance . from this one
would expect the radiative width of the @xmath30 , @xmath31 and @xmath32 to be roughly the same , taking into account merely differences in their @xmath33 , since they are made of quarks with constituent masses of 330450 mev ( depending on their strange quark component ) .
in contrast , experiment gives @xmath34 the recipe for rescuing this potential disaster was suggested long ago by hayne and isgur [ 21 ] .
they proposed multiplying by a factor of the mass of the meson cubed .
this has been justified by ackleh and barnes [ 17 ] by noting that if one considers the point interaction of a pseudoscalar with two photons , then the gauge invariant form of the interaction lagrangian is @xmath35 this gives @xmath36 , where @xmath37 is a pseudoscalar meson and @xmath38 its mass .
if we take it that what we have calculated by considering the fermion exchange graphs of fig . 4a is some effective coupling @xmath39 , then this has to be corrected by @xmath40 , where @xmath41 is typically 1 gev for light hadrons [ 21,16,17 ] .
whilst such factors bring agreement with experiment for the @xmath30 and @xmath31 , they also affect predictions for the @xmath42 , @xmath43 too , because these mass terms appear cubed . in tables
1,2 we show the predictions for radiative widths for different modellings for a handful of charmonium and light quark states .
we will hear at this meeting , new results to compare with these .
the absolute rates for the @xmath18 and @xmath19 look good .
of course , the quark model not only predicts ground states , but radial excitations too . whilst some , like the @xmath44 , have been seen in hadronic reactions for almost twenty years
, it is only very recently that a candidate , the @xmath45 , has been observed in two photon channels by l3 [ 22 ] at lep . for light quark systems , in particular ,
what is needed are calculations beyond the free quark exchange model .
quarks are confined .
there are no poles in their propagators and the bound state dynamics more complicated ( fig .
meaningful non - perturbative computations should be possible soon .
.radiative widths of some charmonium states predicted by variations on the modelling [ 16,17,19 ] described in the text compared with the pdg98 average [ 23 ] in the case of the @xmath46 this is just one experiment . [ cols= " < ,
< , < , < , < " , ] what about the higher mass scalars ? in @xmath47 elastic scattering , the @xmath48 and @xmath49 appear as dips in the cross - section .
it is quite possible that these feed through into the two photon reaction too ( through eq .
if this is the case , then the idea of deducing an upper bound on the radiative width of the @xmath49 by the surplus of events over a smooth background used by aleph [ 43 ] to obtain @xmath50 kev looks highly questionable . only by extending the amplitude analysis
i have just described to higher @xmath47 masses can one be certain of separating an @xmath51wave signal out from under the larger @xmath52wave components and so deduce meaningful results for the radiative widths of the glueball candidates @xmath49 and @xmath53 .
clearly a pure glueball would have a very tiny @xmath3 width ( coming wholly from higher order graphs in perturbative qcd ) .
consequently , its _ stickiness _ ( defined earlier in this section ) would be large .
large _ means has always been problematic .
stickiness [ 32 ] is conventionally normalized so that it is 1 for the @xmath54 , a typical quark model state .
however , for its @xmath55 partner , the @xmath56 , the stickiness is @xmath57 .
clearly , the @xmath20 is not _
sticky _ in the sense of gluey .
this large value happens because its two photon width is small : a result of it being made of charge 1/3 quarks , while the @xmath18 has both @xmath58 and @xmath59 quarks .
the cleo group [ 44 ] have sought to address this problem .
they define a quantity _ gluiness _ , which is roughly equal for the @xmath18 and @xmath20 .
however , gluiness [ 44 ] suffers from the drawback that one has to assume in advance what the charged components of the meson are something , of course , we want to use the two photon width to determine . as already remarked the scalar glueball candidates , primarily the @xmath49 and @xmath53 ,
are unlikely to be pure glue , but inevitably mix with the nearby @xmath8 multiplet(s ) .
as soon as this happens , these states cease to have tiny radiative widths , but have those typical of @xmath8 scalars , cf .
table 4 .
stickiness then has much less relevance .
the branching ratio to glue of akir and farrar [ 36 ] is far more useful , as is the absolute value of their two photon widths .
the predictions for the latter depend very strongly on the assumed mixing scenario as illustrated by comparing the results of close , farrar and li [ 45 ] with those of jaminon and van den bosche [ 46 ] and with burakovsky and page [ 47 ] .
these vary by large factors .
the challenge for the near future is for experiment to deliver accurate measurements , from which the two photon widths of the @xmath49 and @xmath53 can be deduced , and for theory to make reliable predictions for these same states . combining this information we can determine the composition of these key hadrons .
only then will we understand the nature of the light scalar mesons , a nature and composition that is intimately tied to the structure of the qcd vacuum both @xmath8 and glue .
after decades of promise and years of effort , there may be light at the end of the tunnel . | the physics that can be learnt by studying exclusive channels in two photon interactions is recalled .
this serves as an introduction to the exclusive reaction session of photon99 . |
mobile data traffic is predicted to grow by 800% in the next 6 years @xcite .
this imposes a severe strain on existing wireless networks .
one of the most promising methods to reduce downlink traffic is to store content closer to end users , commonly referred to as _ caching _ @xcite .
caching has been studied extensively in recent years . in @xcite , a central server with access to a file library
serves a number of users , each with the ability to cache content , using a common channel .
it is shown that , exploiting local caches as side information , transmitting coded multicasts , i.e. , linear combinations of packets from different files , can significantly reduce the amount of data that is necessary to transmit over the channel .
this concept is investigated in @xcite for the specific application of wireless networks . the central server is omitted and requests for content are served by mobile devices caching data , which broadcast interfile coded messages using device - to - device ( d2d ) communication .
in wireless networks , mobile devices that cache content might move out of range . the impact of data loss when mobile devices leave the network is considered in @xcite .
instantaneous repair of single device departures is investigated in @xcite and scheduled repair of lost content due to multiple devices leaving the network is treated in @xcite and @xcite .
the papers @xcite consider a single file and it is shown that the use of erasure correcting codes help to reduce the amount of data that is downloaded from the base station ( bs ) . in @xcite
, multiple files are cached in a number of small bss from which mobile devices download content .
it is shown that caching content using maximum distance separable ( mds ) codes reduces the macro bs downlink traffic . in this paper
, we consider the use of mds codes to cache content in the mobile devices of a wireless cellular network .
the devices roam between neighboring cells according to a poisson random process , similar to the modeling of device mobility in @xcite , and request files from a library at random times .
files are encoded using mds codes of equal code length but potentially different code rate , and coded symbols are cached in a number of mobile devices .
we formulate a mixed integer linear program ( milp ) to find the mds codes that minimize the network load , assuming a global average cache size constraint ( across all devices ) .
we show that optimizing the rates of the mds codes may yield a significantly lower network load than when caching only the most popular files .
we also show that , for the considered scenario , caching symbols of a given file on all mobile devices , i.e. , maximal spreading @xcite , is optimal .
_ notation : _ @xmath0 and
@xmath1 $ ] denote probability and expectation , respectively . a random variable @xmath2 that follows the binomial distribution with parameters @xmath3 and @xmath4
is denoted by @xmath5 .
we use bold lowercase letters @xmath6 to denote vectors . @xmath7 and @xmath8 are the length-@xmath9 vectors with all entries @xmath10 and @xmath11 , respectively .
we consider a large area of @xmath12 square units ( sq.u . ) consisting of many small cells with equal area @xmath13 sq.u . , each served by a bs .
@xmath14 mobile devices are uniformly spread over the large area , which gives that the number of devices in a given small cell , denoted by @xmath2 , is binomially distributed as @xmath15 the expected number of devices initially available in a small cell , denoted by @xmath16 , is @xmath17 = { m_\text{tot}}\cdot { a}/{a_\text{tot}}.\ ] ] users wish to download files from a library of @xmath18 files that is always available at the bss .
we assume that all files have equal size and that the file popularity follows the zipf distribution @xcite , i.e. , the popularity of file @xmath19 is given by @xmath20 where @xmath21 is the skewness parameter of the distribution . _
content allocation : _ each file @xmath19 that is to be cached is partitioned into @xmath22 packets and encoded into @xmath3 symbols using an @xmath23 mds code @xcite .
the @xmath3 coded symbols are cached in @xmath3 mobile devices ( possibly different for each file ) in the area , chosen uniformly at random .
hence , for each file @xmath19 , the @xmath3 devices caching the file store a fraction @xmath24 of the file .
thus , @xmath25 where @xmath26 implies that file @xmath19 is not cached .
we define the vector @xmath27 and refer to it as the _ content allocation_. the number of caching devices in a given small cell , denoted by @xmath28 , is binomially distributed as @xmath29 and the expected number of caching devices in a given small cell , denoted by @xmath9 , is @xmath30 = n \cdot { a}/{a_\text{tot}}.\ ] ] _ arrival - departure process : _ mobile devices roam in and out of a small cell .
we assume that the arrival of mobile devices caching a coded symbol of a given file and the arrival of devices not caching a symbol of the file are two independent arrival processes with interarrival times exponentially independent , identically distributed ( i.i.d . ) with rates @xmath31 and @xmath32 , respectively , per time unit ( @xmath33 ) .
therefore , the overall arrival rate is @xmath34 .
each device has an exponential i.i.d .
lifetime with rate @xmath35 @xmath33 before departing the small cell .
hence , the number of devices in the cell follows a poisson birth - death process with rate @xmath36 @xcite .
we assume @xmath37 and @xmath38 and , hence , the expected number of devices and the expected number of caching devices in the small cell is @xmath16 and @xmath9 , respectively .
this process can be characterized by an @xmath39 queueing model where the instantaneous probability of having @xmath40 caching devices in the cell is @xcite @xmath41 note that this model predicts a nonzero probability that the number of mobile devices in a small cell is larger than @xmath14 , which violates our initial assumption on the total number of devices in the area .
however , we consider scenarios where @xmath42 , such that this probability is negligibly small . _ data download : _ mobile devices request files at random times with the time between requests exponentially i.i.d . with rate @xmath43 @xmath33
hence , the expected total request rate in the small cell is @xmath44 @xmath33 .
a device request file @xmath19 with probability @xmath45 , given by . due to the mds property ,
@xmath22 coded symbols are sufficient to decode the file @xcite .
the user requesting content downloads as many coded symbols as possible ( up to @xmath22 ) from available caching devices and , if additional symbols are required , these are retrieved from the bs . the equivalent number of files that are downloaded from the bs per t.u .
is defined as the _ downlink rate _ and the equivalent number of files downloaded from caching devices ( per t.u . )
is defined as the _ d2d communication rate_. we assume that the communication is error free and incurs zero delay .
in this section , we derive the probability to have a number of caching devices available at the time of a request . for later use , we define @xmath46 as the event that there are @xmath40 caching devices available at the time of a request for file @xmath19 .
we also define @xmath47 and @xmath48 as the events that the request comes from a caching device and comes from a noncaching device , respectively .
we have the following proposition . @xmath49 the probability of having @xmath40 mobile devices caching a coded symbol of file @xmath19 at the time of a request for file @xmath19 is @xmath50 the probability @xmath51 is given by @xmath52 since the number of caching devices and the number of noncaching devices are poisson distributed with rates @xmath9 and @xmath53 , respectively , the expected fraction of mobile devices caching a coded symbol of file @xmath19 is @xmath54 .
a request is equally likely to come from any mobile device and we have that @xmath55 and @xmath56 using , @xmath57 if there is a request for file @xmath19 from a caching device ( event @xmath47 ) there is at least one caching device in the cell . the probability that there are @xmath40 caching devices in the cell is the probability that @xmath58 additional caching devices are available , i.e. , @xmath59 using . using with in , we get . the amount of data that is downloaded from the bs and from the mobile devices is given by the following proposition .
[ prop : gamma ] provided that there are @xmath40 mobile devices caching a coded symbol of file @xmath19 in the small cell , then , for a given request for file @xmath19 , the fraction of the file that is downloaded from the bs is @xmath60 the fraction of file @xmath19 that is downloaded from caching devices if the request comes from a mobile device not caching a coded symbol of file @xmath19 is @xmath61 and if the request comes from a device caching a symbol of file @xmath19 @xmath62 we require @xmath22 coded symbols to decode file @xmath19 . if at the time of a request for file @xmath19 there are @xmath63 devices caching a symbol of file @xmath19 in the small cell , @xmath64 symbols are retrieved from the bs .
if @xmath65 , no symbols are downloaded from the bs . hence , the fraction of file @xmath19 that is downloaded from the bs is given by .
following the same argument , it is trivial to obtain and .
the expected downlink rate for a content allocation @xmath66 , denoted by @xmath67 , is given by @xmath68 since the amount of data that is downloaded from the bs is the same , irrespective of whether the request comes from a caching or a noncaching device .
the expected d2d communication rate , denoted by @xmath69 , is given by @xmath70 where , using
, , and , @xmath71
we consider the optimization of the content allocation @xmath66 such that the weighted communication rate @xmath72 is minimized . minimizing the expected downlink rate corresponds to @xmath73 .
however , it might be desirable to also limit the d2d communication rate for various reasons , such as device battery constraints and interference between devices . therefore , we consider @xmath74 , where @xmath75 stems from the fact that the bottleneck is in the downlink .
we enforce a global average cache size constraint , denoted by @xmath76 , where @xmath77 this implies an average cache size constraint per device @xmath78 .
the optimization problem can be formulated as [ p1 ] @xmath79 we denote by @xmath80 the _ optimal _ content allocation resulting from .
the objective function can be rewritten as shown in ( at the top of the page ) using and . it is clear that the objective function is a sum of piecewise linear functions of @xmath81 .
this allows us to rewrite the optimization problem in a way that is tractable , using the epigraph formulation @xcite . using and introducing a new optimization variable @xmath82 ,
the optimization problem can be written equivalently as [ p2 ] @xmath83 where the constraints and arise from the first and second term in the @xmath84 function in .
note that we drop the @xmath44 term in since it is irrelevant to the solution of the optimization problem .
the @xmath81 can only take on the discrete values given by . to handle this
, we introduce the binary optimization variable @xmath85 and let @xmath86 such that @xmath87 finally , we formulate an milp that is equivalent to and , minimizing the objective function under the constraints , , and , @xmath88
in figs . [ fig : ratevsn][fig : ratevstheta ] we evaluate the downlink rate and the weighted communication rate for several content allocations @xmath66 . besides the optimal content allocation @xmath80 , we consider a _
popular _ content allocation , where each of the @xmath89 most popular files is cached in @xmath3 ( possibly different ) mobile devices ( i.e. , using an @xmath90 repetition code ) .
the popular content allocation is given by @xmath91 we also consider a _
proportional _ content allocation , @xmath92 where @xmath93 .
we remark that the proportional content allocation typically leads to noninteger @xmath22 file partitions .
albeit impractical , this content allocation is interesting for comparison purposes .
for the results , we consider a file library with @xmath94 files and zipf parameter @xmath95 .
we assume departure rate @xmath96 @xmath33 and request rate @xmath97 @xmath33 .
note that the optimal content allocation does not depend on @xmath43 .
we also assume that the average cache size constraint per device is @xmath98 file .
monte carlo simulation results are also included in the figures .
[ fig : ratevsn ] shows the downlink rate in versus the code length @xmath3 , normalized to the total number of mobile devices in the area @xmath14 , for @xmath99 .
note that @xmath100 corresponds to maximal spreading @xcite , i.e. , storing a coded symbol of a given file on as many devices as possible . in this case
, the global average cache size constraint becomes a strict cache size constraint , i.e. , @xmath101 .
we recall that @xmath102 , using and .
we observe that the downlink rate decreases with @xmath103 for both the optimal and proportional content allocations , i.e. , maximal spreading @xcite is optimal .
this optimality is observed for the weighted communication rate in for several values of @xmath18 , @xmath21 , @xmath16 , @xmath104 , and @xmath105 . for the popular content allocation , the downlink rate increases with @xmath103 .
this is because @xmath76 decreases as @xmath103 increases and a smaller number of complete files can be cached in the mobile devices . for @xmath100 ,
the downlink rate using the popular content allocation is almost 80% higher compared to when using the optimal content allocation . in fig .
[ fig : ratevsm ] , we show the weighted communication rate @xmath106 in versus the expected number of mobile devices in the small cell , @xmath16 , for @xmath100 , i.e. , maximal spreading , and @xmath107 .
the expected accumulated cache capacity of the devices in the small cell , @xmath108 , grows with @xmath16 but so does the expected request rate @xmath44 .
an increase of @xmath108 leads to a decrease of @xmath106 , while an increase of @xmath44 leads to an increase of @xmath106 .
it is observed in the figure that @xmath106 increases with @xmath16 , i.e. , the effect of an increase in weighted communication rate due to the increased request rate dominates the performance . for small @xmath16 , @xmath106 is similar for all considered content allocations .
however , for larger @xmath16 , mds coded caching with optimized code rates yields significantly lower weighted communication rate . compared to the optimal content allocation , the popular content allocation and the proportional content allocation involve roughly a 50% and a 10% increase , respectively , in the weighted communication rate for @xmath109 . in fig .
[ fig : ratevstheta ] , the weighted communication rate in is plotted for the considered content allocations as a function of @xmath105 for @xmath110 and @xmath109 .
we see that the weighted communication rate increases linearly with @xmath105 for the popular and proportional content allocations .
this is expected since these content allocations are obviously not affected by changing @xmath105 .
the weighted communication rate for the optimal content allocation is on the other hand not monotonically increasing with @xmath105 . for @xmath111 ,
the weight of the downlink rate in and the d2d communication rate in is the same , i.e. , it is equally `` expensive '' to download files from the bs and the caching devices .
it is easy to prove that , for integer @xmath76 , the optimal content allocation for @xmath111 is the popular content allocation .
however , this is not the case in general . for the example in fig .
[ fig : ratevstheta ] , @xmath112 , and thus for @xmath111 popular allocation is optimal . for large @xmath105
, the popular content allocation leads to a significant communication overhead compared to the optimal content allocation .
on the other hand , the proportional content allocation incurs a small overhead compared to the optimal content allocation . finally , in fig .
[ fig : allocation ] we show the content allocations as a function of file index used to achieve the weighted communication rate in fig . [
fig : ratevstheta ] for @xmath113 .
we optimized the caching of content in mobile devices using maximum distance separable codes .
we formulated a mixed integer linear program to minimize the weighted sum of the downlink rate and the device - to - device communication rate under a global average cache size constraint .
we showed that optimized mds coded caching may yield a significantly lower weighted communication rate compared to when caching ( uncoded ) the most popular files .
furthermore , our results show that caching coded symbols of a particular file on all devices , i.e. , maximal spreading , is optimal .
j. pedersen , a. graell i amat , i. andriyanova , and f. brnnstrm , `` repair scheduling in wireless distributed storage with d2d communication , '' in _ proc .
theory workshop ( itw ) _ , jeju , korea , 2015 , pp .
6973 .
l. breslau , p. cao , l. fan , g. phillips , and s. shenker , `` web caching and zipf - like distributions : evidence and implications , '' in _ proc .
ieee 18th annu .
joint conf .
ieee comput .
( infocom ) _ , new york , ny , 1999 , pp . | we consider the caching of content in the mobile devices in a wireless network using maximum distance separable ( mds ) codes .
we focus on a small cell , served by a base station ( bs ) , where mobile devices arrive and depart according to a poisson birth - death process .
users requesting a particular file download coded symbols from caching devices using device - to - device communication .
if additional symbols are required to decode the file , these are downloaded from the bs .
we optimize the mds codes to minimize the network load under a global average cache size constraint .
we show that , for most practical scenarios , using optimized mds codes results in significantly lower network load compared to when caching the most popular files .
furthermore , our results show that caching coded symbols of each file on all mobile devices , i.e. , maximal spreading , is optimal . |
.comparison of recent wide - area low - redshift galaxy surveys [ tab : othersurveys ] [ cols="<,^,^,^",options="header " , ] @xmath3 objects removed from the initial target list ( due to changes in + the 2mass source catalogue after 6dfgs was underway ) .
+ i d @xmath4 or 9999 in these cases . +
* columns : * + ( 1 ) i d : programme i d ( progid in the database ; see sec . [
sec : release ] ) .
+ ( 2 ) survey sample : first sample ( in order of progid ) in which + object is found .
+ ( 3 ) 6dfgs spectra : number of spectra obtained for this sample . + note that some objects were observed more than once .
the + numbers include spectra of all qualities and galactic sources .
+ ( 4 ) good @xmath5 : number of robust extragalactic 6dfgs redshifts , + ( those with @xmath6 or 4 ) . reflects contents of database . +
( 5 ) lit .
@xmath5 : additional literature extragalactic redshifts ( ignoring + repeats and overlaps ) .
+ ( 6 ) total @xmath5 : total number of extragalactic redshifts for objects + in this sample .
+ only @xmath7 redshifts should be used in any galaxy analysis .
( the distinction between @xmath8 and @xmath9 is less important than that between @xmath10 and @xmath8 , since the former represent a successful redshift in either case . )
galaxies with repeat observations have all spectra retained in the database , and the final catalogued redshift is a weighted mean of the measurements with @xmath7 , excluding redshift blunders .
descriptions of the redshift quality scheme in its previous forms can be found in sec .
2.1 of @xcite and sec .
4.4 of @xcite .
unreliable ( @xmath10 ) or unusable ( @xmath11 ) galaxy redshifts together comprise around 8 percent of the redshift sample .
galactic sources ( @xmath12 ) represent another 4 percent .
the remaining 110256 sources with @xmath7 are the robust extragalactic 6dfgs redshifts that should be used ( alongside the 14815 literature redshifts ) in any analysis or other application .
tables [ tab : distribution ] and [ tab : breakdown ] give the breakdown of these numbers across various 6dfgs sub - samples .
redshift uncertainties and blunder rates were estimated from the sample of 6dfgs galaxies with repeat redshift measurements . there were 8028 such redshift pairs , 43percent of which were first - year ( pre-2002.5 ) data .
most repeat measurements were made because of a low - quality initial measurement , or because of a change in the field tiling strategy after the first year of observations .
we define a blunder as a redshift mismatch of more than 330 ( @xmath13 ) between a pair of redshift measurements that we would expect to agree .
the blunder rate on individual 6dfgs redshifts is 1.6percent , the same as reported for the first data release @xcite . in late 2002 , a new transmissive volume - phase holographic ( vph ) gratings ( 580v and 425r ) replaced the existing reflection gratings ( 600v and 316r ) , resulting in improved throughput , uniformity , and data quality . excluding first - year repeats reduces the individual blunder rate to 1.2percent .
while the first - year data represent nearly half of all repeat measurements , they represent less than a fifth of the overall survey .
table [ tab : blunder ] summarises the blunder rates and other statistics for both the full and post - first - year data .
figure [ fig:6dfvs6df ] shows repeat redshift measurements for 6dfgs observations with the vph gratings , representative of the great majority of survey spectra ( around 80 percent : 4570 sources spanning 2002.5 to 2006 ) .
blunder measurements ( 106 of them ) have been circled , and the scatter in redshift offset , @xmath14 , as a function of redshift is also shown .
not surprisingly , redshifts becoming increasingly difficult to secure as one moves to higher values .
the inset in figure [ fig:6dfvs6df ] displays the distribution in @xmath15 for measurement pairs grouped by their redshift quality @xmath8 or 4 classifications .
there are 3611 pairs in the non - blunder sample with both measurements of quality @xmath9 , with scatter implying a redshift uncertainty in an individual @xmath9 measurement of @xmath16 .
likewise , the scatter in the much smaller @xmath8 sample ( 33 pairs ) suggests @xmath17 . if we include the pre-2002.5 non - vph data , the implied redshift uncertainties are unchanged for @xmath9 and increase slightly in the case of @xmath8 ( @xmath18 ) . note that these redshift uncertainties are less than those estimated in @xcite from the first data release , demonstrating the improved integrity of the 6dfgs data since the early releases .
llr + * 6dfgs ( full sample ) : * & + total repeat measurements ( @xmath19 ) : & 8028 + rms scatter of all redshift measurement pairs@xmath20 & @xmath21 + @xmath9 redshift uncertainty ( 6051 sources ) & @xmath22 + @xmath8 redshift uncertainty ( 104 sources ) & @xmath23 + & + number of blunders@xmath24 ( @xmath25 ) : & 260 + 6dfgs pair - wise blunder rate : & 3.2% + 6dfgs single - measurement blunder rate : & 1.6% + + * 6dfgs ( vph grating only , 2002.5 2006 ) : * & + total repeat measurements ( @xmath19 ) : & 4570 + rms scatter of all redshift measurement pairs@xmath20 & @xmath23 + @xmath9 redshift uncertainty ( 3611 sources ) & @xmath26 + @xmath8 redshift uncertainty ( 33 sources ) & @xmath27 + & + number of blunders@xmath24 ( @xmath25 ) : & 106 + 6dfgs pair - wise blunder rate : & 2.3% + 6dfgs single - measurement blunder rate : & 1.2% + + * 6dfgs ( vph only ) vs. sdss dr7 : * & + number of comparison sources ( @xmath28 ) : & 2459 + number of blunders@xmath20 ( @xmath28 ) : & 95 + pair - wise blunder rate : & 3.9% + implied blunder rate for sdss : & 2.7% + + @xmath3 clipping the most extreme @xmath29% of outliers ( @xmath30% either side ) .
+ @xmath31 a blunder is defined as having @xmath32(@xmath13 ) .
+ an external comparison of 6dfgs redshifts to those overlapping the seventh data release ( dr7 ) of the sloan digital sky survey ( sdss ; * ? ? ?
* ) was also made and is shown in fig .
[ fig : litvs6df ] .
although the full sdss dr7 contains over a million classified extragalactic spectra , almost all are too northerly to overlap significantly with the southern 6dfgs or are too faint
. however , the 2459 sources in common to both catalogues provide a valuable test of redshift success rates .
the pair - wise blunder fraction in this case is 3.9percent .
splitting this with the 6dfgs blunder rate of 1.2percent implies an sdss blunder rate of 2.7percent , although the 6dfgs blunder rate at the fainter sdss magnitudes is likely to be somewhat higher than the 1.2percent measured overall .
the 6dfgs online database is hosted at the wide field astronomy unit of the institute for astronomy at the university of edinburgh .
data are grouped into 15 inter - linked tables consisting of the master target list , all input catalogues , and their photometry .
users can obtain fits and jpeg files of 6dfgs spectra as well as 2mass and supercosmos postage stamp images in and where available , and a plethora of tabulated values for observational quantities and derived photometric and spectroscopic properties . the database can be queried in either its native structured query language ( sql ) or via an html web - form interface .
more complete descriptions are given elsewhere @xcite , although several new aspects of the database are discussed below .
figure [ fig : example ] shows two examples of the way data are presented in the database .
table [ tab : paramcontents ] shows the full parameter listing for the 6dfgs database .
individual database parameters are grouped into lists of related data called _ tables_. parameter definitions are given in documentation on the database web site .
the target table contains the original target list for 6dfgs , and so contains both observed and unobserved objects .
individual entries in this table are celestial sources , and the targetid parameters are their unique integer identifiers . note that the original target list _ can not _ be used to estimate completeness , due to magnitude revisions in both the 2mass xsc and supercosmos magnitudes subsequent to its compilation .
item ( iv ) below discusses this important issue in more detail .
the spectra table holds the redshift and other spectroscopic data obtained by the 6df instrument through the course of the 6dfgs .
many new parameters have been introduced to this table for this release ( indicated in table [ tab : paramcontents ] by the @xmath3 symbol ) .
individual entries in this table are spectroscopic observations , meaning that there can be multiple entries for a given object .
the specid parameter is the unique integer identifier for 6dfgs observations .
most 6dfgs spectra consist of two halves , observed separately through different gratings , and subsequently spliced together : a v portion ( @xmath335600 ) and an r portion ( @xmath347500 ) . or @xmath35 passbands . ]
( data taken prior to october 2002 used different gratings , spanning 40005600 and 55008400 . ) various parameters in spectra belonging to the individual v or r observations carry a _ v or _ r suffix , and are listed in table [ tab : paramcontents ] for v ( with slanted font to indicated that there is a matching set of r parameters ) .
the twomass and supercos tables hold relevant 2mass xsc and supercosmos photometric and spatial information .
likewise , the remaining eleven tables contain related observables from the input lists contributing additional 6dfgs targets to target . while some of the parameter names have been duplicated between tables ( e.g. mag_1 , mag_2 ) their meaning changes from one table to the next , as indicated in table [ tab : paramcontents ] .
database tables can be queried individually or in pairs .
alternatively , positional cross - matching ( r.a . and dec .
) can be done between database sources and those in a user - supplied list uploaded to the site .
search results can be returned as html - formatted tables , with each entry linking to individual gif frames showing the 6dfgs spectrum alongside its @xmath36 postage stamp images , as shown in fig .
[ fig : example ] .
individual object fits files of the same data can also be accessed in this way .
long database returns can also be emailed to the user as an ascii comma - separated variable ( csv ) text file . alternatively , the fits files of all objects found through a search can be emailed to the user as a single tar file under a _ tar saveset _ option . llcl + table name & description & progid & parameters + + target & the master target list & @xmath37 & targetid , targetname , htmid , ra , dec , cx , cy , cz , gl , gb , + & & & a_v , progid , bmag , rmag , sg , zcatvel , zcaterr , zcatref , + & & & bmagsel , rmagsel , templatecode@xmath3 , framename + spectra & redshifts and observational data & @xmath37 & specid , targetid , targetname , obsra , obsdec , match_dr , + & & & htmid , cx , cy , cz , z_origin , z , z_helio , quality , abemma , + & & & nmbest , ngood , z_emi , q_z_emi , kbestr , r_crcor , z_abs , + & & & q_z_abs , q_final , ialter , z_comm , zemibesterr , zabsbesterr , + & & & zfinalerr , _
title_v , cenra_v , gratslot_v , cendec_v , _ + & & & _ appra_v , appdec_v , actmjd_v , conmjd_v , progid_v , _ + & & & _ label_v , obsid_v , run_v , exp_v , ncomb_v , gratid_v , _ + & & & _ gratset_v , gratblaz_v , source_v , focus_v , tfocus_v , _ + & & & _ gain_v , noise_v , ccd_v , utdate_v , utstrt_v , _ + & & & _ mjdobs_v , name_v , thput_v , ra_v , dec_v , x_v , y_v , _ + & & & _ xerr_v , yerr_v , theta_v , fibre_v , pivot_v , recmag_v , _
+ & & & _ pid_v , _ framename , _
axisstart_v , axisend_v , _ matchspecid , + & & & z_initial@xmath3 , z_helio_initial@xmath3 , z_update_flag@xmath3 , z_update_comm@xmath3 , + & & & slit_vane_corr@xmath3 , quality_initial , xtalkflag@xmath3 , xtalkscore@xmath3 , + & & & xtalkveloff@xmath3 , xtalkcomm@xmath3 , quality_update_comm , deprecated@xmath3 + & & & revtemplate@xmath3 , revcomment@xmath3 , z_comm_inital@xmath3 + twomass & 2mass input catalogues & 1 ( @xmath38 ) , & objid , catname , targetname , targetid , ra , dec , priority , + & & 3 ( @xmath39 ) , & mag_1 , progid , mag_2 , j_m_k20fe , h_m_k20fe , + & & 4 ( @xmath40 ) & k_m_k20fe , radius , a_b , muk20fe , corr , j , h , kext , k , + & & & kext_k , prevcatname@xmath3 , rtot@xmath3 , jtot@xmath3 , htot@xmath3 , ktot@xmath3 + supercos & supercosmos input catalogues & 8 ( ) , & objid , catname , targetname , targetid , ra , dec , priority , + & & 7 ( ) & mag_1 ( old ) , progid , mag_2 ( old ) , comment + fsc & iras faint source catalogue sources & 126 & objid , catname , targetname , targetid , ra , dec , priority , + & & & mag_1 , progid , mag_2 , comment + rass & rosat all - sky survey candidate agn & 113 & objid , catname , targetname , targetid , ra , dec , priority , + & & & mag_1 , progid , mag_2 , comment + hipass & sources from the hipass hi survey & 119 & objid , catname , targetname , targetid , ra , dec , priority , + & & & mag_1 , progid , mag_2 + durukst & durham / ukst galaxy survey extension & 78 & objid , catname , targetname , targetid , ra , dec , priority , + & & & mag_1 , progid , mag_2 + shapley & shapley supercluster galaxies & 90 & objid , catname , targetname , targetid , ra , dec , priority , + & & & mag_1 , progid , mag_2 + denisi & denis survey galaxies , @xmath41 & 6 & objid , catname , targetname , targetid , ra , dec , priority , + & & & mag_1 , progid , mag_2 , comment + denisj & denis survey galaxies , @xmath42 & 5 & objid , catname , targetname , targetid , ra , dec , priority , + & & & mag_1 , progid , mag_2 , comment + agn2mass & 2mass red agn survey candidates & 116 & objid , catname , targetname , targetid , ra , dec , priority , + & & & mag_1 , progid , mag_2 , mag_3 + he s & hamburg / eso survey candidate qsos & 129 & objid , catname , targetname , targetid , ra , dec , priority , + & & & mag_1 , progid , mag_2 + nvss & candidate qsos from nvss & 130 & objid , catname , targetname , targetid , ra , dec , priority , + & & & mag_1 , progid , mag_2 + sumss & bright radio sources from sumss & 125 & objid , catname , targetname , targetid , ra , dec , priority , + & & & mag_1 , progid , mag_2 + + + @xmath3 new parameters created for the final data release .
+ _ slanted font _
v - spectrum parameters ( _ _ v _ ) have matching r - spectrum ( _ _ r _ ) parameters .
+ additional downloads in the form of ascii files are also available from the database web site .
these include a master catalogue compilation of all redshifts ( from both 6dfgs and the literature ) , as well as a comma - separated file of the spectral observations .
the latter contains an entry for every 6dfgs observation held by the database ( including repeats ) , regardless of redshift quality . the master catalogue attempts to assign the best available redshift to those sources determined to be extragalactic . in the case of repeats ,
a combined 6dfgs redshift is obtained by error - weighting ( @xmath43 ) those @xmath7 redshifts within @xmath13 ( 330 ) of an initial @xmath7 median , thereby excluding blunders . where literature redshifts exist and are consistent with the 6dfgs redshift , the latter is used in the catalogue . in cases of disagreement ( @xmath44 difference ) , the 6dfgs redshift is taken and the mismatch is flagged .
literature redshifts are use , where they exist , for objects that 6dfgs failed to secure .
the master catalogue includes the targetid for each object and the specid references for each 6dfgs observation contributing to the final redshift , to facilitate cross - referencing with the 6dfgs database . completeness maps ( calculated from the revised target lists , after 2mass and supercos magnitude changes )
will be made available at a future date .
table [ tab : parameters ] lists a subset of the more commonly - used database parameters , along with detailed descriptions .
new parameters for this final release are indicated .
users should pay particular attention to the important differences between parameters which have similar - sounding names but which are significantly different in purpose .
examples to note are : ( i ) z , z_origin , z_helio , z_initial , and z_helio_initial , ( ii ) quality and q_final , ( iii ) ( jtot , htot , ktot ) , ( j , h , k ) , and ( mag1 , mag2 ) ( from the twomass table ) , and , ( iv ) ( bmag , rmag ) , ( bmagsel , rmagsel ) and ( bmag , rmag ) ( from the supercos table ) . table [ tab : parameters ] details the differences between them . lcl + parameter & associated table(s ) & notes + + & & + targetid & all & unique source i d ( integer ) , used to link tables .
+ targetname & all & source name , ` g@xmath45 ' .
( sources observed but not in the original + & & target list have the form ` c@xmath45 ' ) .
+ progid & all & programme i d ( integer ) , identifying the origin of targets .
@xmath46 for main samples .
+ objid & all except target & unique object i d ( integer ) , assigned to each object in all input catalogues .
+ & and spectra & + & & + bmag , rmag & target & new supercosmos magnitudes following the revision for 2dfgrs by peacock , + & & hambly and read .
first introduced for dr2 . the most reliable 6dfgs magnitudes .
+ zcatvel , zcaterr & target & existing redshifts and errors ( ) from zcat @xcite where available .
+ zcatref & target & code indicating source of zcat redshift : ` @xmath47 ' for earlier 6dfgs redshifts ( subsequently + & & ingested by zcat ) , ` @xmath48 ' for zcat - ingested 2dfgrs redshifts .
the ` @xmath49 ' in both cases + & & holds redshift quality ( see quality below ) . zcatref@xmath50 for other zcat surveys .
+ bmagsel , rmagsel & target & old supercosmos magnitudes compiled by w. saunders .
never used for selection + & & and not intended for science . previously under bmag and rmag in pre - dr2 releases . + templatecode & target & code indicating cross - correlation template : ` n ' = no redshift , ` z ' = zcat redshift + & & ( no template used ) , ` t ' = 2dfgrs ( no template used ) , 1 @xmath51 9 = 6dfgs template code .
+ & & + specid & spectra & unique spectral i d ( integer ) .
different for repeat observations of the same object .
+ z_origin & spectra & is ` c ' for most spectra , which come from ( c)ombined ( spliced ) v and r spectral frames .
+ & & is ` v ' or ` r ' for unpaired ( orphan ) data , as applicable .
+ kbestr & spectra & template spectrum i d ( integer ) used for redshift cross - correlation .
+ z_helio & spectra & heliocentric redshift .
corrected by @xmath52 for template offset if kbestr@xmath53 or 7 .
+ & & the redshift intended for science use .
+ z & spectra & raw measured redshift . not intended for science use . also template offset corrected .
+ z_initial@xmath3 & spectra & initial copy of redshift z , uncorrected ( e.g. for slit vane shifts ) . not for scientific use . + z_update_flag@xmath3 & spectra & z_helio corrections : ` 1 ' if slit - vane corrected , ` 2 ' if template corrected , ` 3 ' for both . + z_helio_initial@xmath3 & spectra & initial version of z_helio , uncorrected ( e.g. for slit vane shifts ) . not for science use .
+ quality & spectra & redshift quality , @xmath54 ( integer ) : ` 1 ' for unusable measurements , ` 2 ' for possible but unlikely + & & redshifts , ` 3 ' for a reliable redshift , ` 4 ' for high - quality redshifts , and ` 6 ' for + & & confirmed galactic sources .
only quality@xmath55 or 4 should be used for science .
( quality + & & does _ not _ measure spectral quality . )
+ q_final & spectra & final redshift quality assigned by software .
not intended for general use .
. + quality_initial@xmath3 & spectra & quality value at initial ingest , before database revision . not for general use
. + quality_update_comm@xmath3 & spectra & explanation of quality value changes during database revision .
+ title_v , title_r & spectra & observation title from sds configuration file ( consisting of field name and plate number ) .
+ xtalkflag@xmath3 & spectra & fibre number of a nearby object suspected of spectral cross - talk contamination . ` @xmath56 ' if + & & object is a contaminator itself .
` 0 ' if neither a contaminator nor contaminee .
+ xtalkscore@xmath3 & spectra & score from ` 0 ' ( none ) to ` 5 ' ( high ) assessing the likelihood of spectral cross - contamination .
+ xtalkveloff@xmath3 & spectra & velocity offset ( ) between contaminator and contaminee in cross - contamination .
+ xtalkcomm@xmath3 & spectra & comment about cross - talk likelihood .
+ slit_vane_corr@xmath3 & spectra & correction ( ) made to a redshift affected by slit vane shifts during observing
. + revtemplate@xmath3 & spectra & code of any spectral template used during the database revision of redshifts
. + revcomment@xmath3 & spectra & explanation of any redshift changes resulting from the database revision .
+ & & + catname & twomass & 2mass name .
( prior to this release , catname held the old names now in prevcatname ) .
+ prevcatname@xmath3 & twomass & old 2mass name ( as at 2001 ) .
+ rtot@xmath3 & twomass & 2mass xsc extrapolated / total radius ( 2mass r_ext parameter ) .
+ jtot , htot , ktot@xmath3 & twomass & revised 2mass xsc total magnitudes ( 2mass j_m_ext , _ etc . _ ) . for science use .
+ mag_1,mag_2 & twomass & input catalogue magnitudes . not used in twomass table and so default non - value is 99.99 .
+ & & superseded by jtot , htot , and ktot .
+ corr & twomass & magnitude correction ( based on average surface brightness ) used to calculate kext_k .
+ j , h , k & twomass & old 2mass xsc total magnitudes .
@xmath57 used for selection .
superseded by jtot , _
etc_. + kext & twomass & redundant 2mass extrapolated @xmath38 magnitudes , previously used to obtain kext_k .
+ kext_k & twomass & old total @xmath38 magnitude estimated from kext and corr . used in original
6dfgs @xmath38-band + & & selection ( see @xcite for a discussion ) . now redundant .
+ & & + mag_1,mag_2 & supercos & old supercosmos magnitudes compiled by saunders , parker and read for target + & & selection .
now superseded by the revised magnitudes bmag and rmag in the target table .
+ & & + + @xmath3 new parameters created for the final data release .
+ all of the changes previously implemented for dr2 @xcite have been retained , with some modifications . in particular ,
some fields rejected from earlier data releases on technical grounds have been fixed and included in the final release .
the final data span observations from 2001 may to 2006 january inclusive .
new changes are as follows : 1 . * revised 2mass names : * between the creation of the initial 6dfgs target list in 2001 and the final 2mass xsc data release in 2004 , the 2mass source designations changed in the last two digits in both the r.a . and dec .
components of the source name .
the original 2mass names ( previously held in the 6dfgs database twomass table under the attribute catname ) have been retained but re - badged under a new attribute prevcatname .
the revised 2mass names are stored in catname and are consistent with the final data release of the 2mass xsc .
original 6dfgs sources that were subsequently omitted from the final 2mass data release have catname= ` ' .
* revised 2mass photometry : * the total magnitudes used to select 6dfgs sources were also revised by 2mass between 2001 and 2004 .
these new values are held in the newly - created jtot , htot , ktot . the revisions amount to less than 0.03mag , except in the case of corrected blunders .
the old magnitudes used for target selection continue to be held in j , h and kext_k , the latter being derived from surface brightness - corrected 2mass extrapolated magnitudes ( see @xcite for a full discussion ) .
* revised supercosmos photometry : * as discussed in @xcite for dr2 , the supercosmos magnitudes were also revised between 2001 and 2004 . as was the case for dr2 , bmag and rmag
are the revised magnitudes , which should be used for science purposes .
however , some of the values in bmag and rmag have changed because of an improvement in the algorithm we have used to match 6dfgs objects with new supercosmos magnitudes .
this has removed much more of the deblending discussed in sec .
2.3 of @xcite .
the historical magnitudes held in bmagsel , rmagsel ( in the target table ) and mag_1 , mag_2 ( in the supercos table ) retain their dr2 definitions and values .
* redshift completeness : * the 2mass and supercosmos magnitude revisions have imparted a small but non - negligible scatter between the old and new versions of , particularly @xmath38 .
they have a non - negligible impact on estimates of 6dfgs redshift completeness at the faint end ( faintest @xmath58 mag ) of each distribution . in this regime ,
the new magnitudes cause increasing numbers of original 6dfgs targets to lie beyond the cut - off and increasing numbers of sources that were not original targets to fall inside the cut - off .
consequently , new target lists were compiled using the revised magnitudes , the completeness estimates were recalculated , and the results are presented along with the luminosity and mass functions in jones et al .
( in prep . ) . 5 .
* fibre cross - talk : * instances of fibre cross - talk , in which bright spectral features from one spectrum overlap with an adjacent one , have been reviewed and are now flagged in the database through three new parameters : xtalkflag , xtalkscore , and xtalkveloff , defined in table [ tab : parameters ] .
the flags are not definitive and are only meant to reflect the _ likelihood _ that a redshift has been affected thus .
specifically , users are urged to use extreme caution with redshifts from sources having @xmath59 , @xmath60 and @xmath61 .
cases of @xmath62 are weak candidates where cross - talk is possible but not fully convincing ( e.g. only the v or the r spectra are affected , but not both ) .
@xmath63 are good candidates , but which carry the previous caveat .
@xmath64 are likely cross - talk pairs which are usually confirmed through visual inspection of the spectra .
cross - talk is an uncommon occurrence ( about @xmath65 percent of all spectra ) , and it only affects the redshifts for spectra with fewer real features than false ones . an algorithm was used to search for coincident emission lines in adjacent spectra and a cross - talk severity value assigned from 1 to 5 .
users are urged to exercise caution with spectra and redshifts having cross - talk values of 3 or greater .
a detailed discussion of the cross - talk phenomenon can be found in the database documentation on the website .
6 . * highest redshift sources : * very occasionally , spurious features due to cross - talk or poor sky - subtraction led to erroneously high redshifts .
this is particularly the case with the additional target samples ( @xmath66 ) , whose selection criteria do not necessarily ensure reliable detections at the optical wavelengths of 6dfgs spectra .
special care should be taken with the high redshift sources reported for these targets .
all sources ( across all programmes ) with @xmath67 were re - examined and re - classified where necessary .
in addition , those sources from the primary and secondary samples ( @xmath68 ) with redshifts in the range @xmath69 were re - examined .
there are 318 6dfgs sources with @xmath70 , mostly qsos , and a further 7 possible cases .
the highest of these is the @xmath71 qso g2037567@xmath37243832 .
other notable examples are the candidate double qso sources g0114547@xmath37181903 ( @xmath72 ) shown in fig .
[ fig : example](@xmath73 ) and g2052000@xmath37500523 ( @xmath74 ) . deep follow - up imaging in search of a foreground source
is necessary to decide whether these sources are individual gravitationally lensed qsos or genuine qso pairs .
even with such data in hand , the distinction is quite often equivocal ( e.g. * ? ? ?
* ; * ? ? ?
* and references therein ) . 7 .
* orphan fields : * the final data release includes ( for the first time ) data from 29 orphan fields .
these are fields that , for various reasons , are missing either the v or r half of the spectrum .
these fields have a reduced redshift yield because of the restricted access to redshifted spectral features , particularly in the case of missing r spectra .
orphan field data are flagged in the database through the z_origin parameter ( see table [ tab : parameters ] ) .
* re - examination of q=1 and q=2 spectra : * all sources originally classified as either being extragalactic and @xmath10 , or non-2mass - selected ( @xmath75 ) and @xmath11 , have been re - examined .
this was done primarily to improve the identification of faint high - redshift qsos .
many qsos were poorly identified in the early stages of the survey due to the absence of suitable qso templates for redshifting .
redshift data for 4506 @xmath10 and 3687 @xmath11 sources were checked , and the database updated where necessary .
* image examination of all q=6 sources and re - redshifting : * in the initial redshifting effort , 6212 sources were classified as @xmath12 ( i.e. confirmed galactic sources with @xmath76 ) on the basis of their spectra and redshifts alone .
once spectral and imaging data were assembled side - by - side in the 6dfgs database , it was straightforward to examine the postage - stamp images of these sources , given their spectral classification .
most were confirmed as being true galactic sources ( stars , hii regions , planetary nebulae , ysos ) , or galactic objects in close proximity to an extragalactic source .
a small number were also found to be 2mass imaging artefacts , or parts of larger objects .
however , a significant number ( 847 ) were found to be galaxies with near - zero redshifts , which were subsequently re - redshifted and re - classified , and updated in the database . in some cases , even though the source was clearly a galaxy on the basis of its imaging , its true redshift could not be obtained .
the most common causes were scattered light from a nearby star , or contamination from a foreground screen of galactic emission . 10 . * anomalous @xmath38@xmath5 sources with q=3,4 : *
the @xmath38@xmath5 magnitude - redshift relation was used to identify anomalous redshifts ( @xmath7 ) outside the envelope normally spanned by this relation at typical 6dfgs redshifts .
the postage - stamp images of these sources were compared to their spectra and redshifts to decide if the initial redshift was incorrect .
there were 120 objects deemed to have an anomalous @xmath38@xmath5 ; 94 were found to have incorrect redshifts , which were re - examined and re - incorporated into the database
* correction of slit - vane shifted fields : * midway through the survey it became apparent that the magnetically - held vane supporting the spectrograph slit was shifting occasionally between exposures .
this problem was discovered prior to dr2 but the affected redshifts were withheld ; they have been corrected and provided in the final release .
the resulting spectra from affected fields show a small wavelength offset ( greater than @xmath77 and up to a few ) , dependent on fibre number .
the v and r spectral halves were sometimes affected individually , and at other times in unison .
instances of shifting were isolated by comparing the wavelength of the [ oi]@xmath785577.4 sky line , as measured from the 6dfgs spectra , to its true value .
a search found 125 affected fields able to be satisfactorily fit ( measured [ oi ] against fibre number ) and redshift corrected . in all , 18438 galaxies were corrected in this way ( approximately 14 percent of the entire sample of _ all _ spectra ) , with corrections @xmath79 .
redshift template values kbestr were used to determine whether to apply a correction .
if an object used @xmath80 ( corresponding to early - type galaxy templates ) , the redshift was deemed to be due to absorption - lines , which occur predominantly in the v half .
if the corresponding v frame was indeed slit - vane affected , a correction was applied to the redshift for this galaxy based on the fit to the v frame _
alone_. alternatively ,
if @xmath81 ( corresponding to late - type galaxy templates ) , then the redshift was deemed to be emission - line dependent , and the corresponding r frame correction was made where necessary .
users can find those galaxies with slit - vane corrected redshifts through the new slitvanecorr parameter , which gives the size ( in ) of any corrections applied .
unaffected galaxies have @xmath82 .
the corrected redshifts are the heliocentric redshifts held by z_helio
correction for template offset values : * various tests comparing 6dfgs redshifts to independent measurements found small systematic offsets in the case of a couple of templates .
the discrepancy is almost certainly due to a zero - point error in the velocity calibration of the template spectra .
this effect was discovered prior to dr2 and is discussed in @xcite , although no corrections were applied to the affected redshifts in that release . for this final release , corrections of @xmath83 have been applied to redshifts derived from templates @xmath84 .
the corrected redshifts are both the raw ( z ) and heliocentric ( z_helio ) redshifts .
the redshift offsets were found to be consistent between a 2004 comparison of 16127 6dfgs and zcat redshifts , and a 2007 comparison of 443 redshifts from various peculiar velocity surveys @xcite . 13 . * telluric sky line subtraction : * the redshifting software used by 6dfgs automatically removed telluric absorption lines from spectra , but
the database spectra have hitherto retained their imprint .
for the final release we have re - spliced spectra and incorporated telluric line removal . an example spectrum is shown in fig . [ fig : example ]
. a small number of spectra which failed to re - splice successfully have had their old telluric - affected versions retained .
* spurious clustering : * the entire sample of reliable redshifts ( @xmath7 ) was tested for spurious clusters , caused by any systematic effect that produces noticeable numbers of objects from the same field with nearly identical redshifts .
possible causes include poor sky subtraction and/or splicing of spectra , and the fibre cross - talk effect discussed in item ( v ) .
fields containing at least 16 cases of galaxy groups ( 3 or more members ) with redshift differences of less than 30 had their redshifts re - examined : 171 galaxies from 7 fields .
no prior knowledge of real galaxy clustering was used for the re - redshifting , and the database was updated with new redshifts and quality assignments .
the field 0058m30 was particularly prominent with 48 galaxies at or near an apparent redshift of 0.1590 .
this was due to the over - subtraction and subsequent mis - identification of the 7600 telluric absorption band with redshifted h@xmath85 .
a further 134 objects with redshifts in the range @xmath86 were reexamined for this effect , and 118 given corrected @xmath5 or @xmath54 values .
almost all of the affected spectra are among the earliest observations of survey data ( 2001 ) , prior to the switch to vph gratings
* rass sources : * all sources in the rosat all - sky survey ( rass ) additional target sample ( @xmath87 ; 1850 sources ) were re - examined using the full qso template set .
the database was updated with new redshifts and quality assignments .
@xcite describe the selection and characteristics of this sample in more detail .
the wide sky coverage of the 6df galaxy survey affords the most detailed view yet of southern large - scale structures out to @xmath8830000 .
the 6dfgs extends the sky coverage of the 2dfgrs @xcite by an order of magnitude , and likewise improves by an order of magnitude on the sampling density of the all - sky pscz survey @xcite .
prominent southern structures such as shapley , hydra - centaurus and horologium - reticulum have received much special attention in their own right over recent years @xcite . however , a detailed large - scale mapping of all intervening structures ( and the voids between them ) with a purpose - built instrument has remained unavailable until now . the complementary 2mass redshift survey ( 2mrs ; huchra et al .
, in prep ) uses the 6dfgs in the south to provide an all - sky redshift survey of some 23000 galaxies to @xmath89 ( @xmath90 ) .
it is hoped it will one day be extended to reach an equivalent depth to 6dfgs in the north in those areas not already covered by sdss . figures [ fig : colplot1 ] and [ fig : colplot2 ] show the @xmath91 universe as seen by 6dfgs in the plane of the sky , projected in galactic coordinates .
the two figures show the northern and southern galactic hemispheres , respectively .
familiar large - scale concentrations such as shapley are obvious , and several of the key structures have been labelled . at @xmath92 ,
filamentary structures such as the centaurus , fornax and sculptor walls @xcite interconnect their namesake clusters in a manner typical of large structures generally . at @xmath93 to 0.01 the centaurus wall crosses the galactic plane zone of avoidance ( zoa ) and meets the hydra wall at the centaurus cluster .
the hydra wall then extends roughly parallel to the zoa before separating into two distinct filaments at the adjacent hydra / antlia clusters , both of which extend into the zoa . behind these , at @xmath94 to 0.02 , a separate filament incorporates the norma and centaurus - crux clusters , and encompasses the putative great attractor region ( * ? ? ?
* ; * ? ? ?
* and references therein )
. beyond these , at @xmath95 to 0.05 , lies the shapley supercluster complex , a massive concentration of clusters thought to be responsible for 10 percent of the local group motion @xcite or even more @xcite .
figures [ fig : pieinner ] and [ fig : pieouter ] show an alternative projection of these structures , as conventional radial redshift maps , cross - sectioned in declination .
the two figures show the same data on two different scales , out to limiting redshifts of @xmath96 and 0.1 respectively .
the empty sectors in our maps correspond to the zoa region .
these declination - slice sky views can also be cross - referenced with the aitoff - projected sky redshift maps presented in @xcite for the 6dfgs data available up to 2004 , as well as figs [ fig : colplot1 ] and [ fig : colplot2 ] .
figure [ fig : pieouter ] similarly displays the local universe out to @xmath97 with hitherto unseen detail and sky coverage .
while it extends and confirms the now familiar labyrinth of filaments and voids , it also reveals evidence of inhomogeneity on a still larger scale the plot for @xmath98 ( middle right panel ) is a good example .
a large under - dense region ( @xmath99 ) at @xmath100hr to 5hr separates regions of compact high - density filaments ; similar inhomogeneities are visible in the other plots .
an extraordinarily large void ( @xmath101 by 0.07 ) is apparent in the plot for @xmath102 , towards @xmath103hr .
other voids of this size are apparent when the data are examined in cartesian coordinates .
the most extreme inhomogeneity , however , is the over - dense shapley region , which is unique within the sample volume .
@xcite have used spherical harmonics and wiener filtering to decompose the density and velocity field of the shallower 2mrs .
the correspondence between the largest - scale superclusters and voids seen in both surveys at @xmath104 is clear .
our southernmost projection ( @xmath105 ) confirms the most distant ( pavo ) of the three tentative superclusters of @xcite while indicating that the other two are not major overdensities .
we point out that this southern region is where 6dfgs coverage is generally lowest , with below - average completeness between 0 hr and 6 hr and around the pole ( poor sky coverage ) , and at 11 hr to 17 hr ( zoa ) .
azimuthal stretching effects are also evident , due to the wide r.a .
span of single fields at polar declinations .
work is currently underway cataloguing new clusters and groups from 6dfgs ( merson et al , in prep . )
using a percolation - inferred friends - of - friends algorithm @xcite . at the same time
, a preliminary list of @xmath106 void regions has been compiled as a reference for future work on under - dense regions .
a power spectrum analysis of the clustering of 6dfgs galaxies will be published elsewhere .
the 6df galaxy survey ( 6dfgs ) is a combined redshift and peculiar velocity survey over most of the southern sky . here
we present the final redshift catalogue for the survey ( version 1.0 ) , consisting of 125071 extragalactic redshifts over the whole southern sky with @xmath0 . of these
, 110256 are new redshifts from 136304 spectra obtained with the united kingdom schmidt telescope ( ukst ) between 2001 may and 2006 january . with a median redshift of @xmath107
, 6dfgs is the deepest hemispheric redshift survey to date .
redshifts and associated spectra are available through a fully - searchable online sql database , interlinked with photometric and imaging data from the 2mass xsc , supercosmos and a dozen other input catalogues .
peculiar velocities and distances for the brightest 10 percent of the sample will be made available in a separate future release . in this paper
we have mapped the large - scale structures of the local ( @xmath108 ) southern universe in unprecedented detail .
in addition to encompassing well - known superclusters such as shapley and hydra - centaurus , the 6dfgs data reveal a wealth of new intervening structures .
the greater depth and sampling density of 6dfgs compared to earlier surveys of equivalent sky coverage has confirmed hundreds of voids and furnished first redshifts for around 400 southern abell clusters @xcite .
more detailed quantitative analyses of 6dfgs large - scale structure will be the subject of future publications .
the unprecedented combination of angular coverage and depth in 6dfgs offers the best chance yet to minimise systematics in the determination of the luminosity and stellar mass functions of low - redshift galaxies , both in the near - infrared and optical ( e.g. * ? ? ?
while surveys containing @xmath109-galaxy redshifts ( such as 6dfgs ) have now reduced random errors to comparable levels of high precision , systematic errors remain the dominant source of the differences between surveys .
for example , the evolutionary corrections that initially beset comparisons between 2dfgrs and sdss ( cf . * ? ? ?
* ; * ? ? ?
* ) are negligible for 6dfgs , which spans lookback times of only 0.2 to 0.7
gyr across @xmath110 $ ] ( compared to 0.5 to 1.3 gyr for sdss and 2dfgrs ) .
the minimisation of such systematics are a feature of the 6dfgs stellar mass and luminosity functions derived for the final redshift set ( jones et al .
, in prep ) .
in addition to these studies , 6dfgs redshift data have already been used to support a variety of extragalactic samples selected from across the electromagnetic spectrum .
deep hi surveys planned for next - generation radio telescopes @xcite will also benefit from this redshift information as they probe the gas content of the local southern universe over comparable volumes .
dhj acknowledges support from australian research council discovery projects grant ( dp-0208876 ) , administered by the australian national university .
jph acknowledges support from the us national science foundation under grant ast0406906 .
we dedicate this paper to two colleagues who made important contributions to the 6df galaxy survey before their passing : john dawe ( 1942 2004 ) , observer and long - time proponent of wide - field fibre spectroscopy on the ukst from its earliest days , and tony fairall ( 1943 2008 ) , whose unique insights from a career - long dedication to mapping the southern universe underpin much of the interpretation contained herein . , h. , tago , e. , einasto , m. , einasto , j. , jaaniste , j. , 2005 , in nearby large - scale structures and the zone of avoidance , astronomical society of the pacific conference series vol .
329 , a. p. fairall and p. a. woudt eds .
, p283 , f. g. , parker , q. a. , bogatu , g. , farrell , t. j. , hingley , b. e. , miziarski , s. , 2000 , in iye m. , moorwood a. f. , eds .
spie vol . 4008 , optical and ir telescope instrumentation and detectors , spie , bellingham , wa , p. 123 | we report the final redshift release of the 6df galaxy survey , a combined redshift and peculiar velocity survey over the southern sky ( @xmath0 ) .
its 136304 spectra have yielded 110256 new extragalactic redshifts and a new catalogue of 125071 galaxies making near - complete samples with @xmath1 .
the median redshift of the survey is 0.053 .
survey data , including images , spectra , photometry and redshifts , are available through an online database .
we describe changes to the information in the database since earlier interim data releases .
future releases will include velocity dispersions , distances and peculiar velocities for the brightest early - type galaxies , comprising about 10% of the sample .
here we provide redshift maps of the southern local universe with @xmath2 , showing nearby large - scale structures in hitherto unseen detail .
a number of regions known previously to have a paucity of galaxies are confirmed as significantly underdense regions .
the url of the 6dfgs database is http://www-wfau.roe.ac.uk/6dfgs
. surveys galaxies : clustering galaxies : distances and redshifts cosmology : observations cosmology : large scale structure of universe |
the potential speedup offered by quantum computers is exemplified by shor s factoring algorithm @xcite , grover s search algorithm @xcite , and algorithms for quantum simulation @xcite .
although the origin of this speed - up is not fully understood , there are indications that quantum entanglement plays a crucial role in making quantum algorithms efficient @xcite .
in particular , it was shown that quantum algorithms that do not create entanglement can be simulated efficiently on a classical computer @xcite .
it is therefore of interest to quantify the entanglement produced by quantum algorithms and examine its correlation with their efficiency .
this requires to develop entanglement measures for the quantum states of multiple qubits that appear in quantum algorithms .
these include pure states as well as mixed states , which would inevitably show up when decoherence effects are taken into account .
the special case of bi - partite entanglement has been studied extensively in recent years and suitable entanglement measures were introduced .
it was established that bi - partite entanglement can be considered as a resource for teleportation @xcite .
the entanglement of bi - partite pure states can be evaluated by the von neumann entropy of the reduced density matrix , traced over one of the parties .
for bi - partite mixed states , several measures were proposed @xcite and for the special case of states of two qubits an exact formula for the entanglement of formation was obtained @xcite . for mixed states of multiple qubits , entanglement measures based on distance measures in hilbert space were proposed @xcite .
consider a mixed quantum state @xmath2 of @xmath0 qubits .
the state is non - entangled , or separable , if its density matrix can be written in the form @xmath3 where @xmath4 , @xmath5 is a density operator of a pure state of the @xmath6th qubits , namely @xmath7 and @xmath8 . in the special case
that @xmath2 is a pure state , all probabilities vanish except for @xmath9 , and the state can be expressed by @xmath10 such states are called tensor - product states . in order to evaluate the entanglement of a quantum state , @xmath2 , one needs a scalar function @xmath11 [ or @xmath12 for pure states ] called an _ entanglement measure _ that satisfies @xcite : ( a ) @xmath13 if and only if @xmath2 is a separable state ; ( b ) assuming that each qubit is held by a different party , it is not possible to increase @xmath11 by local operations and classical communication ( locc ) between the parties .
consider the special case of local unitary operators .
such operators can not decrease @xmath11 because if they could then the inverse operators ( which are also unitary ) would increase it and thus contradict the second condition above .
the conclusion is that local unitary operators can not change @xmath11 .
the groverian entanglement measure , @xmath14 , provides an operational measure of entanglement for pure states of multiple qubits @xcite .
it is related to the success probability of grover s search algotirhm when the state @xmath15 is used as the initial state .
a pre - precessing stage is allowed in which an arbitrary local unitary operator is applied to each qubit .
these operators are optimized in order to obtain the maximal success probability of the algorithm , @xmath16 .
the groverian measure is given by @xmath17 .
the groverian measure was used in order to evaluate the entanglement in certain quantum states of high symmetry as well as in states that are generated during the operation of quantum algorithms @xcite .
for example , it was found that grover s iterations generate highly entangled intermediate states , even in case that the initial and the final states are product states . in this paper
we generalize the groverian entanglement measure to the case of mixed states . the groverian measure , @xmath18 , of a given mixed state @xmath2 , of @xmath0 qubits , is obtained by its purification into a pure state of @xmath1 qubits . an optimization procedure based on uhlmann s theorem @xcite
is then applied before the resulting pure state is fed into grover s algorithm .
@xmath18 is then expressed in terms of the maximal success probability @xmath19 , as described above for pure states . in sec .
[ sec : algorithm ] we briefly describe grover s search algorithm , in a context suitable to this paper . in sec .
[ sec : pure ] we review the groverian entanglement measure for pure states .
the generalization to mixed states is presented in sec .
[ sec : mixed ] and its operational interpretation is considered .
the results are summarized and discussed in sec .
[ sec : summary ] .
consider a search space @xmath20 containing @xmath21 elements .
we assume , for convenience , that @xmath22 , where @xmath0 is an integer .
this way , the elements of @xmath20 can be represented by an @xmath0-qubit register @xmath23 , with the computational basis states @xmath24 , @xmath25 . we assume that one element in the search space is marked , namely it is the solution of the search problem .
the distinction between the marked and unmarked elements is expressed by a suitable function , @xmath26 , such that @xmath27 for the marked element , and @xmath28 for the rest .
the search for the marked element now becomes a search for the element for which @xmath27 . to solve this problem on a classical computer one needs to evaluate @xmath29 for each element , one by one , until a marked state is found .
thus , on average , @xmath30 evaluations of @xmath29 are required and @xmath21 in the worst case . for a quantum computer , on which @xmath29 is evaluated _ coherently _ , it was shown that a sequence of unitary operations , called grover s algorithm and denoted by @xmath31 , can locate the marked element using only @xmath32 coherent queries of @xmath29 @xcite . starting with the equal superposition state , @xmath33 and applying the operator @xmath31 one obtains @xmath34 , where @xmath35 is the marked state .
thus , the success probability of the algorithm is almost unity @xcite . with this performance ,
grover s algorithm was shown to be optimal @xcite namely , it is as efficient as theoretically possible @xcite .
the adjoint equation takes the form @xmath36 where the error is due to the discreteness of the grover iterations @xcite .
if an arbitrary pure state , @xmath37 , is used as the initial state instead of the state @xmath38 , the success probability is reduced @xcite .
it is given by @xcite @xmath39 using eq .
( [ eq:<mu ] ) we obtain @xmath40 namely , the success probability is determined by the overlap between @xmath15 and @xmath38 .
consider grover s search algorithm , in which an arbitrary pure state @xmath15 is used as the initial state . before applying the operator @xmath31
, there is a pre - processing stage in which arbitrary local unitary operators , @xmath41 , @xmath42 , @xmath43 , @xmath44 , are applied on the @xmath0 qubits in the register ( fig .
[ fig1 ] ) .
these operators are chosen such that the success probability of the algorithm would be maximized .
the maximal success probability is thus given by @xmath45 using eq .
( [ eq:<mu ] ) , this can be re - written as @xmath46 or @xmath47 , where @xmath48 is the space of all tensor product states of the form @xmath49 .
the groverian measure is given by @xcite @xmath50 for the case of pure states , for which @xmath14 is defined , it is closely related to an entanglement measure introduced in refs .
@xcite and was shown to be an entanglement monotone .
the latter measure is defined for both pure and mixed states .
it can be interpreted as the distance between the given state and the nearest separable state and expressed in terms of the fidelity of the two states . based on these results
, it was shown @xcite that @xmath14 satisfies : ( a ) @xmath51 , with equality only when @xmath37 is a product state ; ( b ) @xmath14 can not be increased using local operations and classical communication ( locc ) . therefore , @xmath14 is an entanglement monotone for pure states .
a related result was obtained in ref .
@xcite , where it was shown that the evolution of the quantum state during the iteration of grover s algorithm corresponds to the shortest path in hilbert space using a suitable metric . .
an arbitrary pure state @xmath15 of @xmath0 qubits is inserted as the input state . in the pre - processing stage ,
a local unitary operator is applied to each qubit before the resulting state is fed into grover s algorithm .
the local unitary operators @xmath52 , @xmath53 are optimized in order to maximize the success probability of the search algorithm for the given initial state @xmath15 . , width=302 ]
the quantum circuit that demonstrates the evaluation of the groverian measure @xmath18 for a mixed state @xmath2 is shown in fig .
the given state @xmath2 is purified into a pure state @xmath37 of @xmath1 qubits . in the pre - processing an operator @xmath54
is applied .
this operator has the property that its adjoint operator satisfies @xmath55 , where @xmath56 is a purification of a separable state of @xmath0 qubits .
the resulting state is then inserted into grover s search algorithm .
an optimization is performed over the entire family of operators @xmath54 that satisfy the above condition , in order to maximize the success probablity of the search algorithm .
equivalently , the optimization can be performed over all the separable states , @xmath57 , and over all possible purifications @xmath56 of each one of them .
the maximal success probability for the initial state @xmath2 is denoted by @xmath19 and the groverian measure is given by @xmath58 .
the construction of the groverian measure for mixed states is based on a purification process , where the pure state @xmath37 of @xmath1 qubits is a purification of the mixed state @xmath2 of @xmath0 qubits .
similarly , @xmath59 is a purification of a separable mixed state @xmath57 . as in the case of pure states ,
we introduce a pre - processing stage before the state is inserted into grover s algorithm . during pre - processing an optimization
is performed over a certain class of unitary operators @xmath54 .
these operators satisfy @xmath60 , where @xmath59 is a purification of a separable state @xmath57 of @xmath0 qubits .
the maximal success probability is given by @xmath61 using eq .
( [ eq:<mu ] ) we obtain @xmath62 where @xmath63 is the set of separable states of @xmath0 qubits .
the maximization is over all separable states @xmath57 of @xmath0 qubits , and for each of them , over all possible purifications @xmath56 of @xmath1 qubits .
this can be rewritten as @xmath64 the first maximization is over all possible states @xmath59 of @xmath1 qubits which are purifications of separable states @xmath57 of @xmath0 qubits .
the second maximization is over all states @xmath15 of @xmath1 qubits which are purifications of @xmath2 .
according to uhlmann s theorem , the fidelity of any two states @xmath2 and @xmath57 of @xmath0 qubits satisfies latexmath:[\[f(\rho,\sigma ) = \max_{|\phi\rangle}\max_{|\psi\rangle } where @xmath59 and @xmath37 , of @xmath1 qubits , are purifications of @xmath2 and @xmath57 respectively @xcite . a useful corollary ( presented in excercise 9.15 on page 411 of ref .
@xcite ) enables us to remove the optimization on @xmath66 , leading to @xmath67 where @xmath37 is an arbitrary purification of @xmath2 . using this corollary we find that @xmath68 an entanglement measure for mixed states , @xmath2 , should satisfy the conditions : ( a ) @xmath69 , where the equality is obtained only if @xmath2 is a separable state ; ( b )
@xmath18 can not be increased using locc . to express the second condition
we introduce a complete set of operators , defined by @xmath70 where @xmath71 and @xmath72 . in this notation
, the condition is that for every density matrix @xmath2 @xmath73 we proceed to prove that @xmath18 fulfills the two conditions described above : as shown in refs .
@xcite , functions of the form of @xmath14 satisfy the conditions above for an entanglement monotone .
more specifically , there exist two specific purifications , @xmath74 and @xmath75 , for which @xmath76 .
thus , @xmath77 if and only if @xmath78 . in this case @xmath79 , or @xmath80 , thus @xmath81
. since @xmath57 is separable then so is @xmath2 , and the first condition is satisfied . in order to prove the second condition one can use the monotonous quality of the fidelity under trace preserving operations @xcite
this means that for every complete set of operators @xmath82 where the separable state @xmath57 remains separable under the transformation . as a result , @xmath83 therefore
, @xmath18 can not increase under such transformations . of a mixed state of @xmath0 qubits .
the given state @xmath2 is purified into a state @xmath15 of @xmath1 qubits . in the pre - processing stage ,
a unitary operator @xmath84 is applied on @xmath15 , before the resulting state is fed into grover s search algorithm in the space of @xmath1 qubits .
the operator @xmath84 belongs to a class of operators that their adjoint operators satisfy @xmath85 where @xmath56 is a purification of a separable state of @xmath0 qubits .
the choice of @xmath84 is optimized in order to maximize the success probability of the search process for the given state @xmath2 .
, width=302 ] consider a mixed state @xmath57 of @xmath0 qubits .
the state @xmath57 can be purified into a pure state of @xmath1 qubits , half of them associated with the original subspace @xmath86 and the rest with the added subspace @xmath87 @xcite .
such purification takes the form @xmath88 where @xmath89 and @xmath90 are orthonormal basis states and @xmath91 and @xmath92 are unitary operators , in @xmath87 and @xmath86 , respectively . for any separable mixed state ,
@xmath57 of @xmath0 qubits , the unitary operators @xmath91 and @xmath92 and the operator @xmath93 provide a convenient parametrization of the pure state @xmath59 .
this follows from the fact that eq .
( [ eq : separable ] ) can also be expressed as @xmath94 , where @xmath95 , @xmath5 is the density operator of a mixed state of the @xmath6th qubit . to obtain the operator @xmath93 , one needs to diagonalize @xmath57 , according to @xmath96 , where @xmath97 is a suitable unitary operator , resulting in a diagonal matrix @xmath20 .
taking its square root one obtains another diagonal matrix @xmath98 with matrix elements @xmath99 . the process is completed by @xmath100 .
the operator @xmath93 is not a required to be unitary and is not used in the quantum circuit . in order to complete the construction of the quantum circuit
, one needs the operator @xmath101 that transforms @xmath38 into a purification @xmath56 of a separable state @xmath57 . to this end
we construct two bases of the space spanned by @xmath1 qubits
. the first basis is obtained from the computational basis , @xmath102 by appling the hadamard transform on all qubits , namely @xmath103 .
note that @xmath104 .
the second basis , @xmath105 , can be constructed using the gram - schmidt algorithm , starting with @xmath106 . the unitary operator @xmath107 transforms the state @xmath38 into the state @xmath56 , which is a purification of a separable state , @xmath57 , of @xmath0 qubits .
the groverian entanglement measure for multiple qubits , previously introduced for pure states , was generalized to the case of mixed quantum states .
this generalization provides an operational measure of entanglement for both pure and mixed states .
the operational interpretation is based on the fact that the groverian measure of a state @xmath2 is related to the success probability of grover s search algorithm when @xmath2 is used as the initial state , following a suitable pre - processing stage .
when @xmath2 is a mixed state , the pre - processing includes a purification procedure followed by the application of a certain unitary operator . in case
that the given state @xmath2 is a pure state , @xmath18 coincides with the groverian measure for pure states , introduced in ref .
@xcite .
it might seem surprising that in order to evaluate the entanglement of an @xmath0-qubit system one needs a search space of @xmath1 qubits .
this can be explained by considering mixed states as open systems , where , in some cases , the mixture represents an effective entanglement with external qubits .
purification of the mixed state @xmath2 to @xmath1 qubits enables us to create a closed system with no entanglement to any external qubits , in which all the relevant information is maintained .
the groverian measure for mixed states enables to fully capture the entanglement in @xmath2 .
mathematically , this measure satisfies the conditions required from an entanglement measure .
in particular , it vanishes for any separable mixed state of @xmath0 qubits , it is invariant under local unitaries and monotonically decreasing ( in the weak sense ) under locc .
quantum algorithms are designed to start with a well defined initial state .
the final state , just before the measurement is taken , can be either a basis state or a superposition of basis states .
for example , in grover s algorithm with a single marked state , the desired final state ( namely the marked state ) is a basis state . in grover s algorithm with several marked states , as well as in shor s factoring algorithm ,
the desired final state is a superposition .
the analysis presented in this paper can be generalized by replacing grover s algorithm by some other quantum algorithm .
if in the replacement algorithm the desired final state is a basis state , @xmath18 will not depend on the specific algorithm .
however , for algorithms such as shor s algorithm , the maximal success probability may not coincide with the expression used in the groverian measure . yet , in the special case of grover s algorithm with several marked states , the groverian measure still holds @xcite .
recently , the groverian measure was applied in order to evaluate the entanglement in certain pure quantum states of multiple qubits @xcite .
a convenient parametrization was developed that enables analytical calculations of @xmath14 for some pure states of high symmetry . in order to evaluate it for arbitrary states ,
a numerical minimization procedure , based on the steepest descent algorithm was developed . using this procedure ,
the entanglement of intermediate states , generated during the evolution of grover s algorithm , was calculated .
it was found that even if the initial state and the target state are product states , in intermediate stages of the algorithm , highly entangled states are generated , in agreement with earlier studies in which other measures were used @xcite .
this result is interesting in the context of attempts to examine the role of entanglement in quantum algorithms and specifically in grover s search algorithm . in particular
, recent studies have shown that an implementation of grover s algorithm using classical media , namely , in which quantum entanglement does not play a role , would require an exponentially larger overhead compared to the quantum case @xcite .
unlike the case of pure states of two qubits , in which the von neumann entropy provides a complete characterization of the entanglement , multiple qubit states support a large number of different measures @xcite .
it seems that the issue of what measure is relevant depends on the specific physical or operational context in which it is used . in particular , the groverian entanglement measure is motivated by a quantum algorithm .
it thus appears to be a suitable measure for the evaluation of the entanglement that is produced during the evolution of quantum algorithms .
the actual evaluation of entanglement measures turns out to be a difficult computational problem .
this is due to the fact that these measures are typically defined as an extremum of some multi - variable function .
a singular result in this context is the explicit formula for the entanglement of formation of mixed states of two qubits , obtained in refs .
@xcite .
a related operational interpretation of the fidelity , which is also based on uhlmann s theorem , was introduced in ref .
@xcite . in that case
the fidelity @xmath108 of the output states of a noisy channel provides an upper bound on the overlap of the input states , under the assumption that they were pure .
the generalization of the groverian measure to mixed states may provide further insight into the role of entanglement in making quantum algorithms powerful
. it would be interesting to use this measure to evaluate the entanglement generated by quantum algorithms using mixed states , particularly when decoherence effects are taken into account . , generalized grover search algorithm for arbitrary initial amplitude distribution , proceedings of the 1st nasa international conference on quantum computing and quantum communications , palm springs , california , usa , february 17 - 20 , 1998 , lecture notes in computer science 1509 , p. 140
, edited by c.p
. williams ( springer - verlag , 1998 ) . . | the groverian entanglement measure introduced earlier for pure quantum states [ o. biham , m.a .
nielsen and t. osborne , phys .
rev .
a 65 , 062312 ( 2002 ) ] is generalized to the case of mixed states , in a way that maintains its operational interpretation .
the groverian measure of a mixed state of @xmath0 qubits is obtained by a purification procedure into a pure state of @xmath1 qubits , followed by an optimization process before the resulting state is fed into grover s search algorithm .
the groverian measure , expressed in terms of the maximal success probability of the algorithm , provides an operational measure of entanglement of both pure and mixed quantum states of multiple qubits .
these results may provide further insight into the role of entanglement in making quantum algorithms powerful . |
radio sources constitute an unequally powerful tool to investigate the origin and evolution of our universe .
in fact , radio emission is not obscured or attenuated by dust or gas , so deep radio surveys offer a unique opportunity to detect and study sources up to the highest redshifts .
it is following this rationale that very ambitious programs like the planned square kilometer array ( ska , carilli et al . , 2004 )
facility , a gigantic radio telescope with a total collective area of 1 square kilometer , or its precursors askap ( australian ska pathfinder , johnston s. et al .
2007 ) and meerkat ( jonas j.l . 2009 ) will soon see their first light .
when applied to radio sources , the expected advantages in sensitivity , field - of - view , frequency range and spectral resolution will yield substantial progress in many research fields , from cosmology to astrophysics .
extra - galactic radio sources are however a mixed bag of astrophysical objects : amongst the most powerful ones we find radio - loud qsos and frii ( fanaroff & riley 1974 ) galaxies while , moving to weaker sources , the dominant populations become fri , low - excitation galaxies and star - forming galaxies , whose contribution to the total radio counts become predominant at the sub - mjy level ( see e.g. magliocchetti et al . 2000 ; prandoni et al .
discerning amongst these objects and their relative weight is an impossible task to carry out only on the basis of radio - continuum data .
this is why increasingly more effort was recently put on photometric and spectroscopic follow - up studies of radio sources detected in deep - field surveys ( amongst the many : schinnerer et al . 2004 ; 2007 ; 2010 for the cosmos field ; morrison et al . 2010 for goods - n ;
mc alpine et al .
2013 for the video - xmm3 field ; bondi et al .
2003 for the vvds field : simpson et al .
2012 for the subaru / xmm - newton deep field : mao et al .
2011 for the extended chandra deep field ) .
investigations of the optical properties of radio sources can then disentangle the different populations and also allow the determination of a number of quantities which influence the source behavior like e.g. the black - hole mass in optical spectra dominated by broad emission lines ( e.g. metcalf & magliocchetti 2006 ) .
following the launch of the _ spitzer _ satellite , in recent years radio sources have also been investigated at mid - infrared ( mir ) wavelengths ( e.g. appleton et al .
2004 ; boyle et al .
2007 ; magliocchetti , andreani & zwaan 2008 ; garn et al . 2009 ;
leipski et al . 2009 ; de breuck et al .
2010 ; norris et al .
2011 amongst the many ) .
these studies proved to be extremely useful for two main reasons : they allowed investigations of the properties and evolution of the sub - population of star - forming galaxies up to @xmath8 , while on the other hand they provided interesting hints on the central engine responsible for radio ( and mir ) agn emission .
it is only in these very last few years that the radio community has started to realize the importance of multi - wavelength studies which also include far - infrared ( fir ) information .
this growing interest is primarily due to the advent and launch of the _ herschel _ satellite , which for the first time has observed galaxies at fir wavelengths up to very large cosmological ( @xmath9 ) distances ( lutz 2014 ) .
earlier fir missions such as iras , iso and _
spitzer_@70@xmath10 m only probed the relatively local ( @xmath11 ) universe , and since agn - powered radio galaxies in the local universe are almost always hosted by massive red galaxies with little or no ongoing star - formation activity ( e.g. magliocchetti et al . 2002 ; 2004 ) , and since fir emission in galaxies is almost entirely due to processes connected with ongoing star formation , it follows that any pre-_herschel _ interest in a joint radio - fir analysis was limited to re - assessing the precision of the tight radio - fir correlation found by early iras - based studies of local star - forming galaxies ( condon et al . 1982 ) .
however , when moving to higher redshifts , galaxies undergo dramatic changes .
for instance , it is now well established that both the cosmic star - forming activity and agn output steadily increase with look - back time at least up to redshifts @xmath12 ( e.g. gruppioni et al .
2013 : merloni , rudnick & di matteo 2004 ) , and many galaxies are observed to host both an active agn and ongoing star - formation ( e.g. alexander et al .
it is then legitimate to wonder whether radio galaxies also follow such a trend and if in the early universe they are also associated with ongoing stellar production .
first attempts in this direction were presented by the work of seymour et al .
( 2011 ) which analyses the fir emission as observed by the spire instrument ( griffin et al .
2010 ) on board of the _ herschel _ satellite ( pilbratt et al .
2010 ) for a sample of very powerful , l@xmath13 w hz@xmath5 radio sources selected in the extra - galactic _
first look survey field , and by the work of del moro et al .
( 2013 ) who investigate hidden agn activity in a sample of goods - north sources with deep radio , infrared and x - ray data .
magliocchetti et al .
( 2014 ) pushed the seymour et al .
( 2011 ) results further and investigated , for the first time , the fir properties of radio - selected agn of _ all _ radio luminosities and at _ all _ redshifts @xmath14 .
this was done by using the very deep radio catalogue obtained at 1.4 ghz in the cosmos field by schinnerer et al . 2004 ; 2007 ; 2010 and
bondi et al . (
2008 ) ; agn were selected solely on the basis of their radio - luminosity and fir information came from the pep survey ( lutz et al .
results from this analysis indicate that the probability for a radio - selected agn to be detected at fir wavelengths is both a function of radio power and redshift .
powerful sources were found more likely to be fir emitters at earlier epochs because of two distinct effects : 1 ) at all radio luminosities , fir activity monotonically increases with look - back time and 2 ) in the earlier universe , radio activity of agn origin is increasingly less effective at inhibiting fir emission .
magliocchetti et al .
( 2014 ) also found that the observed fir ( and mir ) emission was due to processes which were indistinguishable from those which power star - forming galaxies . the above results ,
although interesting and obtained on a relatively large area , however need the complementarity of deep enough fir observations which can match the depth of the radio data . as a matter of fact , only @xmath15% of the sources identified in the cosmos field as radio - emitting agn possessed a counterpart in the pep catalogues either at 100 @xmath10 m or at 160 @xmath10 m .
the present work then overcomes the above limitations and analyses the fir properties of radio - selected agn on three fields where fir observations are the deepest - to - date : goods - north , goods - south and the lockman hole .
fir data come from both the pep survey ( lutz et al .
2011 ) and the _ herschel_-goods survey ( elbaz et al .
2011 ) and , in the best - observed areas such as the two goods fields , ensure limiting fir fluxes which are deeper by as much as a factor @xmath16 at 100 @xmath10 m and @xmath17 at 160 @xmath10 m than the pep observations on cosmos . throughout the work we will assume a @xmath18cdm cosmology with @xmath19 ( @xmath20 ) , @xmath21 , @xmath22 and @xmath23 . in all cases , masses for the sources under examination are calculated by using the bruzual & charlot ( 2003 ) templates and adopting a salpeter ( 1955 ) initial mass function ( imf ) .
] the radio sources used in this work have been taken from the paper of ibar et al .
( 2009 ) . for consistency with the other surveys considered in this paper , we focussed on the catalogue of objects observed with the very large array ( vla ) at 1.4 ghz .
the sensitivity of these observations was @xmath24jy beam@xmath5 , with a resolution ( full width half maximum - fwhm ) of @xmath17 arcsec .
ibar et al .
( 2009 ) provide a catalogue of 1303 unique sources ( 1346 including sub - components of composite radio objects ) , selected to have a peak - to - local noise ( pnr ) greater than 5 .
as these objects were also observed at 610 mhz with the giant metrewave radio telescope ( gmrt ) , the overwhelming majority of them is also endowed with an estimate of the radio spectral index @xmath25 ( hereafter just @xmath26 for clarity , where @xmath26 is defined as f@xmath27 , with f radio flux ) .
ibar et al .
( 2009 ) also provide a classification based on the radio source appearance , and distinguish them into single ( s ) , double ( d ) , triple ( t ) and extended ( m ) , plus some more complicated morphological cases such as single source with one close neighbour ( sd ) , single source with multiple close neighbours ( st ) and so on .
figure [ fig : lh ] shows the radio sources as taken from the ibar et al .
( 2009 ) 1.4 ghz selection as ( blue ) crosses .
optical , near - ir and mid - ir information for these sources , as well as their redshift determination , can be obtained from the work of fotopoulou et al .
( 2012 ) which provides very precise broad band photometry and photometric redshifts for about 188,000 sources located in @xmath28 deg@xmath29 of the lockman hole area ( green small dots in figure [ fig : lh ] ) .
special care was taken for objects identified as agn .
note that the fotopoulou et al .
( 2012 ) catalogue also includes 388 x - ray detected sources ( the overwhelming majority of which are active galactic nuclei ) identified in the deep xmm - newton observations by brunner et al .
( 2008 ) , available for an area of 0.2 deg@xmath29 .
+ optical identifications for the ibar et al . ( 2009 ) radio sources were then obtained from the fotopoulou et al .
( 2012 ) catalogue for a maximum tolerance radius of 0.9@xmath30 .
this value was chosen so to maximize the number of real associations , while reducing the percentage of spurious matches to roughly 6@xmath31 . the total number of radio sources on the area covered by the broad - band observations of fotopoulou et al .
( 2012 ) is 1309 .
the number of sources with a photometric i d obtained in the way described above is 723 , corresponding to 55,7% of the original parent sample . ]
the fraction of radio sources in the lockman hole endowed with either a spectroscopic or photometric redshift determination as a function of integrated 1.4 ghz radio flux is shown in figure [ fig : fraction_lh ] by the red circles , where the associated error - bars correspond to poissonian estimates . from the figure , one can notice that such a fraction is constant within the errors , and equal to approximately 55% of the parent sample throughout the whole flux range probed by the ibar et al .
( 2009 ) survey .
this indicates that no radio - flux bias affects the optical identifications of the radio sources under consideration .
a smaller area , corresponding to roughly 24@xmath32 , has also been observed by the pacs instrument ( poglitsch et al .
2010 ) onboard of the _ herschel _ satellite ( pilbratt et al .
2010 ) as a part of the pacs evolutionary probe ( pep , d. lutz et al .
2011 ) survey , aimed at studying the properties and cosmological evolution of the infrared population up to redshifts @xmath33 .
1328 sources were observed in this field at either 100 @xmath10 m or 160 @xmath10 m by the pep survey and the blind ( i.e. source extraction performed without relying on priors ) catalogues include objects down to a 3@xmath34 confidence level respectively of @xmath35 mjy and @xmath36 mjy . these are shown in figure [ fig : lh ] by the ( magenta ) open triangles .
698 radio sources from the ibar et al .
( 2009 ) catalogue reside within the area covered by the pep observations , corresponding to 53,8% of the original sample .
+ in order to provide the radio sources with a fir counterpart , we followed the approach presented in magliocchetti et al .
( 2014 ) and cross - matched the radio and fir catalogues with a maximum coherence radius of 4 arcsecs .
this choice insures a negligible contribution from spurious associations ( @xmath37% ) .
the number of radio sources with a fir counterpart at either 100 @xmath10 m or 160 @xmath10 m is 422 , corresponding to 60% of the parent radio catalogue .
the blue squares in figure [ fig : fraction_lh ] represent the trend for fir identifications as a function of 1.4 ghz radio flux .
this percentage is roughly constant up to about 100@xmath10jy and then slowly starts declining beyond that value .
the green triangles in figure [ fig : fraction_lh ] represent the fraction of radio sources which are endowed with both a redshift determination and a fir counterpart .
such a fraction is roughly equal to 40% at the lowest fluxes and then also starts declining beyond radio fluxes f@xmath38jy . ]
deep , 1.4 ghz , radio observations of the whole extended chandra deep field south ( ecdfs ) are presented in miller et al .
this survey covers about a third of a square degree and images 883 radio sources out of which 17 are likely multiple - component objects at the 5@xmath34 level down to a peak rms sensitivity of 6 @xmath10jy .
a smaller area of the ecdfs , corresponding to what is called the goods southern field ( hereafter goods - s ) , is one of the best studied regions in cosmology .
thanks to the very deep and exquisite observations taken at all wavelengths , santini et al .
( 2009 ) were able to provide a high quality photometric catalogue which covers all available wavelengths from 0.3 to 24 micron .
the goods - music catalogue ( santini et al . 2009 ) also includes high precision photometric redshifts for about 15,000 objects and , whenever possible , spectroscopic redshifts for about 3000 of them .
we then cross - correlated the miller et al .
( 2013 ) radio catalogue with the goods - music one in order to provide the 1.4 ghz - selected radio sources with a redshift determination .
142 radio objects out of the the 883 included in the miller et al .
( 2013 ) catalogue fall within the region covered by the goods - music observations .
the distribution of the miller et al .
( 2013 ) 1.4 ghz - selected sources on the goods - s field is shown in figure [ fig : gs ] by the ( blue ) crosses .
the green - shaded area instead shows the distribution of sources from the goods - music catalogue . ] once again with the aim of maximizing the number of sources with a redshift determination , while at the same time minimizing the number of spurious matches , we chose the value of 0.8@xmath30 as the maximum radius to be considered for radio - to - optical associations . this procedure leads to 111 photometrically identified radio sources , corresponding to @xmath39% of the parent catalogue .
the percentage of expected spurious associations is limited to a mere 3.5% .
+ the fraction of radio sources in the goods - s endowed with a redshift determination as a function of integrated 1.4 ghz radio flux is shown in figure [ fig : fraction_gs ] by the red circles , where the associated error - bars correspond to poissonian estimates . from the figure , one can notice that such a fraction is constant within the errors , and equal approximately to 80% of the parent sample throughout the whole flux range probed by the miller et al .
( 2013 ) survey . as it was in the case of the lockman - hole
, the above trend proves that no radio - flux bias affects the optical identifications of the radio sources in exam .
far infrared information for the goods - s radio sources can be obtained through the pep ( lutz et al .
2011 ) and pep / goods - h catalogues ( magnelli et al .
this last one combines observations of the goods fields from the pep and goods-_herschel _ ( elbaz et al .
2011 ) key programs .
this catalogue reaches 3@xmath34 depth of 0.9 , 0.6 and 1.3 mjy respectively at 70 , 100 and 160 @xmath10 m and includes 1635 sources . as the original pep catalogue for goods - s covers a wider area than that of pep / goods - h , we have identified the fir counterparts in two steps , first by using the pep catalogue , and then the pep / goods - h one .
as already specified in the previous section , the maximum tolerance radius chosen to associate a radio source to a fir counterpart was 4@xmath30 .
this procedure leads to 115 , 1.4 ghz - selected , radio sources with a fir counterpart ( corresponding to @xmath40% of the parent catalogue ) , out of which 110 come from the pep catalogues either at 100 or at 160 @xmath10 m and 5 from the combined pep / goods - h catalogue .
the expected fraction of random associations is less than 2% .
+ the trend for the fraction of fir identifications as a function of integrated radio flux is presented in figure [ fig : fraction_gs ] by the blue squares , while that for sources with both a fir counterpart and a redshift determination by the green triangles .
errors correspond to poissonian estimates .
note that in both cases the fractions tend to stay constant , equal to the values of @xmath41% and @xmath42% up to radio fluxes f@xmath43jy , then they start to monotonically decline . ]
deep , 1.4 ghz , radio observations of the goods - north field ( hereafter goods - n ) are presented in morrison et al .
these authors provide a catalogue of 1230 discrete radio sources , imaged over an area of 40@xmath44 .
the catalogue comprises all sources above a 5@xmath34 detection threshold of @xmath45 @xmath10jy at the field centre .
+ the goods - n , likewise the goods - s , is provided with extremely deep observations at all wavelengths .
this enables to derive photometric redshifts for the majority of the sources residing there .
furthermore , many spectroscopic redshifts are also available , and the pep collaboration ( berta et al . [email protected] ] ) provides a list of sources observed in the optical and nir bands , the majority of which are listed with a redshift , stellar mass and age determination .
we then cross - correlated the goods - n radio - catalogue with that of optically - selected sources presented above .
the number of radio objects which fall within the area observed at the other wavebands is 401 . out of these , 295 have a counterpart in the photo - z catalogue , out to a maximum radius of 0.7 arcsecs .
267 of them , corresponding to @xmath47 of the parent radio catalogue , are also endowed with a redshift determination . ]
as it was for goods - s , far infrared information for the goods - n radio sources were obtained through the pep ( lutz et al . 2011 ) and pep / goods - h ( magnelli et al . 2013 ) catalogues . once again
, the maximum tolerance radius chosen to associate a radio source to a fir counterpart was 4@xmath30 .
this procedure leads to 315 , 1.4 ghz - selected , radio sources with a fir counterpart either at 100 @xmath10 m or at 160 @xmath10 m .
this corresponds to @xmath48% of the parent catalogue , in good agreement with what found in the goods - south .
also in this case , the expected fraction of random associations is less than 2% .
+ the trend for the fraction of fir identifications as a function of integrated radio flux is presented in figure [ fig : fraction_gn ] by the blue squares , while that for sources with a redshift determination by the red circles and that for sources with both a fir counterpart and a redshift determination by the green triangles . in all cases errors correspond to poissonian estimates .
note that , once again , while the fraction of sources with a redshift determination tends not to depend on flux at all radio fluxes , the fraction of radio sources with a fir counterpart stays constant , equal to the values of @xmath41% , only up to radio fluxes f@xmath49 @xmath10jy . beyond this value
such a ratio starts to monotonically decline .
the same trend is also reflected in the behavior of the fraction of radio sources with both a redshift determination and a fir counterpart .
table 1 presents a summary of the properties of the 1.4 ghz - selected radio sources on all the three fields considered in this work .
.properties of the analysed samples .
the various columns represent the three different fields considered in this work .
the first row indicates the total number of radio sources selected at 1.4 ghz on regions of the sky covered by deep _
herschel_/pacs observations , n@xmath50 , the second one the number of radio sources with either a spectroscopic or a photometric redshift determination , n@xmath51 , the third row the number of radio sources with a fir counterpart either at 100 @xmath10 m or at 160 @xmath10 m , n@xmath52 , and the last row the number of radio sources with both a fir counterpart and a redshift determination , n@xmath53 . [ cols="<,<,<,<",options="header " , ] some interesting pieces of information on the general properties of radio - emitting agn can then be gathered through investigation of figures [ fig : agnvsflux ] and [ fig : agnvsz ] . in all considered fields their redshift distributions , which we remind , due to the exquisite depths of the radio surveys considered in this work ,
refer to _ complete agn samples selected in radio luminosity _
, present a prominent peak at around @xmath54 and then a possible secondary bump for @xmath8 .
this is in agreement with the results found by magliocchetti et al .
( 2014 ) for the cosmos field .
furthermore , the redshift distribution of agn which also present fir emission closely follows that of their parent population .
once again this is true for all the considered fields , as well as it was in cosmos ( magliocchetti et al .
it follows that , in all cases , the fraction of fir emitters stays constant throughout the whole redshift range up to the earliest epochs probed by our analysis ( @xmath33 ) , i.e. _ there is no preferred age for radio - emitting agn to be associated with fir emission_. indeed , the only appreciable difference amongst the different fields is the value of such a fraction , which is lower and equal to approximately 50% in the lockman hole , while reaches values of about 70 - 80% in the two goods fields where , we remind , fir observations are the deepest ones available to date .
a point which is worth stressing is the extremely high number of radio - emitting agn which are found to possess a counterpart in _ herschel _ maps .
this can be appreciated both in table 2 and in figures [ fig : agnvsflux ] and [ fig : agnvsz ] .
in fact , as already mentioned , one goes from the @xmath2% of the lockman hole , to the @xmath41% of the goods fields .
these figures have to be compared with the @xmath55% found in cosmos ( magliocchetti et al .
2014 ) , where the different percentages are only due to the different depths of fir observations on the different fields . allowing for the relatively large uncertainties due to the small areas covered , these results imply that the overwhelming majority ( @xmath56% , percentage which rises to @xmath57% in the goods - n ) of radio - emitting agn in the goods - s and goods - n are imaged at fir wavelengths down to fluxes 0.6 mjy at 100 @xmath10 m and 1.3 mjy at 160 @xmath10 m .
this is a rather surprising result , since at least locally agn selected at radio wavelengths have always being associated with passive galaxies , with little or no ongoing stellar activity .
another remarkable point that emerges from our analysis is that deeper fir observations do not pick up any new population nor they do single out any different property of the considered population .
in fact , the only effect of increased depth is that of evenly increasing the number of sources which are also fir emitters , independent of look - back time , radio flux and ( as it will be more clear later on ) many other physical parameters which characterize the sources under examination .
] important information on the properties of the inner engine of radio - emitting agn and assessment on whether there are any differences in the presence of fir emission can be gathered by investigation of the distributions of radio luminosities and radio spectral indices for radio - emitting agn both in the presence and in absence of fir emission .
the top panels of figure [ fig : agnvsp ] illustrate the distribution of agn as a function of radio luminosity in the four fields considered here ( lockman hole , goods - s , goods - n and the combined goods - s and goods - n ) .
once again , the solid lines represent the full population of radio - emitting agn , while the dashed ones show the subsets which also have counterparts at fir wavelengths .
the bottom panels instead present the distribution of the fraction of fir emitters as a function of radio luminosity .
in all the considered fields , there is a systematic drop of sources beyond luminosities p@xmath58 w hz@xmath5 sr@xmath5 .
this is due to the fact that very bright radio - selected agn are extremely rare , and finding them in small fields like those considered here is rather unlikely .
once again , the relative fraction of fir emitters varies according to the depth of fir observations on the field , and goes from the @xmath59 50% of the lockman hole to the @xmath60 % of the goods - n .
some interesting information on the behavior of such sources can be found when splitting the samples into different redshift bins .
this is done in figure [ fig : pdistrib_z ] for both the lockman hole and the combined goods field . despite the relative small number of sources
, one can clearly see that the number of p@xmath61 w hz@xmath5 sr@xmath5 radio - emitting agn with a fir counterpart steadily increases as one moves back to the earlier universe . in more detail , for the combined goods fields
we find that _ all sources _ with p@xmath58 w hz@xmath5 sr@xmath5 and @xmath6 have a counterpart at fir wavelengths and , at the same redshifts , between 70% and 80% of them are imaged in _
maps in the luminosity range @xmath62 w hz@xmath5 sr@xmath63 p@xmath64 w hz@xmath5 sr@xmath5 .
the case for the lockman hole is substantially similar , even though the fraction of fir emitters is not as high as in the goods fields as in this latter case fir observations are shallower .
a better way to visualize the above trend is provided by figure [ fig : agnvsp_z ] which shows the fraction of fir emitting agn as a function of radio luminosity in the same three different redshift bins as in figure [ fig : pdistrib_z ] . to increase the statistical significance of the results
, we have plotted this quantity for all the three fields combined together . in this case
, one can appreciate that for redshifts greater than @xmath65 there is a substantial constancy with radio luminosity of the fraction of agn which are also fir emitters .
in other words , _ radio - selected agn at redshifts @xmath7 have all the same chances of being fir emitters independent of their radio luminosity_. however , this is not true at lower redshifts .
in fact the plot shows that in the more local universe there is a ( @xmath66 ) trend for fir emitters to be mostly associated to low - luminosity sources .
such a probability then decreases for brighter agn .
we further stress that , despite the different depth of fir observations , the same behaviors are observed both in the lockman hole and in the combined goods fields , therefore strengthening the significance of our statement .
what one can conclude from this analysis is that chances for an agn which is active at radio wavelengths to also emit at fir wavelengths is maximum at high ( @xmath6 ) redshifts . for deep enough fir observations such as those performed on the goods fields , this probability reaches the value of @xmath59 100% , independent of radio luminosity .
this implies that _ almost all _ radio - selected agn at @xmath6 are also the site of fir emission .
the situation changes in the more local , @xmath11 , universe , as in this latter case one finds that the relative fraction of fir - active agn is lower and , most importantly , decreases as a function of radio luminosity .
in other words , one has that in the local universe , mostly low - luminosity agn are found to be associated with fir emission
. only very few higher - power , p@xmath61 w hz@xmath5 sr@xmath5 , sources instead appear on fir maps .
we note that this conclusions , although based on a relatively small sample of sources agree and further extend to fainter fir emission those obtained by magliocchetti et al .
( 2014 ) for the cosmos field which , although covering a wider area , suffered of the limits due to relatively shallow fir observations .
they also agree with the local results which find radio - emitting agn to reside in early - type galaxies , with little or no on - site ongoing star - forming activity .
our data however indicate that this is only true in the local universe , while in the more distant one fir emission is extremely common amongst radio - active agn . further information on the engine of radio - active agn can be obtained by investigating the spectral indices of their radio emission . as already discussed in 2.1 ,
the lockman hole has been observed by ibar et al .
( 2009 ) both at 1400 and at 600 mhz .
most of the observed sources ( 41 out of 45 ) are endowed with solid estimates of the quantity @xmath67 .
the trends for this quantity as a function of both redshift and radio luminosity are shown in figure [ fig : alpha_agn ] .
squares represent the entire population , filled ones are for the sub - set of fir emitters .
the horizontal dotted line shows the average @xmath68 value obtained for the entire population , the dashed one represents that of @xmath69 derived for fir emitters .
as it is clear also from these visual representations , there is no specific trend in the distribution of @xmath26 values due to the eventual presence of fir emission within the galaxies host of a radio - active agn .
all the distributions look identical , irrespective of the fact that the radio - active agn also show emission at fir wavelengths .
this strongly suggests that , at least at radio wavelengths , the central engine responsible for the agn phenomenon is oblivious to the fir activity which is ongoing within the same host galaxy or , in other words , that fir processes do not influence the activity of the central engine , at least that detected at radio wavelengths .
] ] the results obtained in 4 have clearly shown that fir activity is a very common process amongst radio - active agn . in fact , up to @xmath60% of the radio agn population is found to show up in the _ herschel _ maps .
we have also shown that , except maybe for the most local ( @xmath70 , cfr .
figure [ fig : agnvsz ] ) universe , the probability for a radio - active agn to also emit at fir wavelengths is independent of redshift .
this implies that there is no preferred epoch for a galaxy to be the site of both radio - agn and fir activities .
furthermore , we have found that , at high enough ( @xmath6 ) redshifts , such a probability does not even depend on the power of radio emission from the central black hole .
radio luminosity of agn origin only starts becoming important in the low - redshift universe , whereby the chances of finding fir emission within the host galaxy are maximum for low - luminosity sources , while they monotonically decrease for increasing radio luminosities .
perhaps more importantly , our study has proven that there is no difference between the ( radio ) agn emission associated to galaxies with ongoing fir activity and that in those galaxies which are fir - quiet .
all the above conclusions strongly suggest that , at least for what concerns radio emission , the central engine ignores the existence of fir activity within the same host galaxy , both in terms of its cosmological evolution , of its inner properties and , except for in the very local universe , also in terms of its radiative power . in other words , fir activity does not influence radio agn activity and except for in the very local universe the two processes seem to proceed independently of each other . but what happens at the host galaxy level ?
is there any difference in the properties of galaxies which only host a radio - active agn or are the sites of both radio agn and fir activity ? in order to answer these questions ,
we have then looked into the distributions of stellar masses and ages for galaxies hosts of the agn phenomenon only and of both agn and fir activity .
this has only been possible for the goods - n and goods - s fields , as the lockman hole does not have estimates for either masses or ages of the galaxies within the field .
the distribution of stellar masses for galaxies in the combined goods - s and goods - n fields host of the radio - active agn selected in our work is presented in the top - left panel of figure [ fig : mass_gs ] .
the solid histogram refers to the whole population , while the dashed histogram indicates the sub - class of fir emitters .
the bottom - left panel shows the ratio between these two quantities .
two features can be clearly appreciated from these plots : the first one is that agn hosts are very massive galaxies .
indeed , the mass distribution in both fields peaks in the range m@xmath71 m@xmath4 , and more than 95% of the objects have masses larger than @xmath72 m@xmath4 . furthermore , a non negligible fraction of these sources ( @xmath73 % ) possesses stellar masses in the range @xmath74 $ ] m@xmath4 .
+ the second feature that can be appreciated is that there is no difference in the mass distribution of fir - active and fir - inactive agn . as clearly shown in the bottom - left panel of figure [ fig : mass_gs ] , the two distributions are identical . not only that .
such a constancy is seen throughout the whole redshift range probed by our data ( cfr .
right - hand plot of figure [ fig : mass_gs ] ) , implying that there was never any difference between the masses of the hosts of agn phenomena alone and the masses of the hosts of both agn and fir activity .
this finding is somehow in contradiction with the results of magliocchetti et al .
( 2014 ) on the cosmos field , which indicated that fir emitters preferred somehow slightly lower - mass galaxies ( log@xmath75[m@xmath76m@xmath4]-log@xmath75[m@xmath78m@xmath79\sim$ ] 0.5 ) .
however , the discrepancy can be explained by recalling that cosmos is covered by much shallower fir observations than the two goods fields and therefore the information one can gather from its analysis is possibly affected by biases due to incompleteness .
indeed , only @xmath80% of the radio - active agn in cosmos were found to have a counterpart in _ herschel _ maps , while this percentage rises to @xmath60% in the goods fields .
some more information on these galaxies can be obtained , albeit only in the case of the goods - n , from investigations of the distribution of their ages , @xmath81 .
this is presented in figure [ fig : age ] , which shows as a solid histogram the distribution of ages for the entire agn population , and with the dashed one that for the sub - population of fir emitters .
although the statistical significance of the result ( @xmath82 ) is hampered by the small number of sources , figure [ fig : age ] suggests that even in this case the presence of fir emission does not alter the age properties of the galaxies under exam , at least up to about 2 - 3 gyr . for galaxies older than this value
, instead there seems to be a deficit of fir emitters .
the bulk of the distribution , both in the presence and absence of fir emission , is found for relatively young , @xmath83 gyr sources . beyond this value ,
the number of agn drops dramatically , as only @xmath84% of them is older than this value . on the basis of what found in this section
, we can then conclude that neither the age , nor the mass distribution of galaxies host of a radio agn phenomenon vary in the presence of fir emission .
magliocchetti et al . ( 2014 ) have proven that fir emission from the hosts of radio agn was entirely to be attributed to star - forming processes within the galaxy itself .
this is also true in the present case .
figure [ fig : q100_160 ] reports the distribution of fir , f@xmath85/f@xmath86 , colours as a function of redshift and for different ranges of radio luminosity .
the figure refers to all the three considered fields , whereby small symbols are for goods - s , intermediate - size ones for goods - n and large symbols are for the lockman hole .
the dashed and dotted lines represent the spectral energy distributions ( sed ) of two well studied templates for star - forming galaxies such as m82 ( dashed line ) and apr220 ( dotted line ) .
it is clear that , irrespective of cosmic age and radio luminosity , and except for very few outliers , the fir colour distribution of radio - active agn perfectly follows those of typical moderate - to - intense star - forming galaxies .
the radio excess due to the presence of a radio - active agn is only visible , and not in all cases , in the distributions of the quantities q@xmath87=log [ f@xmath88/f@xmath89 and q@xmath90=log [ f@xmath91/f@xmath89 represented in figure [ fig : q160 ] .
so , we have shown that fir emission from radio - selected agn originates from star - forming processes within the host galaxy . but how bright are these sources at fir wavelengths ? and what are their star - formation rates ? in order to provide such answers ,
we have then estimated the total ir luminosities , l@xmath52 , of those radio - active agn which showed a counterpart in _ herschel _ maps , and subsequently calculated their star - forming rates ( sfr ) according to the standard relation ( kennicut 1998 , which holds for a salpeter imf and for stellar masses in the range @xmath92 m@xmath4 ) : sfr [ m@xmath4 yr@xmath5]=@xmath94 l@xmath52/l@xmath95 .
these are shown in figure [ fig : lir ] , where the left - hand plot refers to the lockman hole , while the right - hand one is for sources in the combined goods - n+goods - s field .
the bottom x - axes report the bolometric luminosities , while sfrs are shown on the top x - axes .
note that the three objects in the combined goods - n+goods - s field endowed with bolometric luminosities above @xmath96 l@xmath4 have estimated redshifts between @xmath97 and @xmath98 . until such redshifts are spectroscopically confirmed , caution should be taken in interpreting results connected to these galaxies .
our data show how radio - active agn are extremely luminous sources : their bolometric luminosities extend up to @xmath99 l@xmath4 , which in turn correspond to extremely high , sfr @xmath100 m@xmath4 yr@xmath5 , star forming rates .
these results imply that galaxy hosts of radio agn activity are indeed also the sites of extremely intense star forming emission .
agn activity does not seem to inhibit star formation within the host galaxy , just as fir activity does not seem to affect radio agn luminosity , at least in the relatively distant ( @xmath6 , cfr 4 ) universe .
in other words , during most of the life - time of such sources , agn and star formation do not seem to influence the activity of each other .
furthermore , as we learned in 5 , also the physical properties of the galaxies host of a radio - active agn are only if any very minimally affected by the presence of a star - forming event within the same galaxy . given the extremely high rate of associations which , we re - iterate , approaches @xmath60% in deep fields such as the goods - n , we stress that the above conclusions do not simply refer to a limited sub - sample of radio - active agn but rather include the overwhelming majority of the considered sources : most of radio - active agn are associated to intense episodes of star - formation
. however , the two processes proceed independently within the same galaxy , at all redshifts except in the most local universe where , possibly intense radio activity acquires enough power to switch off the on - site star formation .
in the previous sections we have shown that there is substantially no difference between the global properties of radio - selected agn and those of the sub - population of sources which are also associated with episodes of star - formation within their hosts .
these findings suggest that , at least in galaxies hosting a radio - active agn event , star - formation is not the main trigger responsible for shaping the formation and evolution of such sources .
what we are left with is then the presence of the agn .
is this really the main factor which influences galaxy evolution ? in order to investigate this issue in more detail , we have compared the distributions of masses and infrared luminosities / star - formation rates for the hosts of radio - selected agn with those of radio - selected star - forming galaxies , derived from the samples considered in the present work on the basis of their radio luminosity as explained in 3 .
this is shown in figure [ fig : mass_lir ] , where the filled squares refer to agn , while the open circles to star - forming galaxies .
the left - hand side plot shows the results obtained for the combined goodsn and goodss fields .
the right - hand side plot instead refers to cosmos , as the lockman hole does not have estimates for the masses of the galaxies which reside there ( cfr .
5 ) . objects in the cosmos field belong to the sample presented by magliocchetti et al .
( 2014 ) and have been selected in the same way as sources in the present work . in order to guide the eye ,
the dashed lines in both plots represent the so - called main sequence for star - forming galaxies obtained by rodighiero et al .
( 2011 ) for z @xmath101 galaxies , while the dotted line shows that derived in the local universe by brinchmann et al .
( 2004 ) and peng et al .
( 2010 ) .
figure [ fig : mass_lir ] clearly shows that the distribution of radio - selected sources which contain a central radio - active agn is markedly different from that of those galaxies which do not .
first of all , agn are much more fir - bright objects .
the overwhelming majority of these sources indeed exhibits star formation rates above @xmath102 m@xmath4yr@xmath5 , while star - forming galaxies are also found down to very low ir luminosities ( of the order of a few @xmath103 l@xmath4 ) . at the same time
, the luminosity distribution of radio - selected agn extends up to extremely bright luminosities , both in the case of cosmos and in the goodss+goodss fields , while that of star - forming galaxies stops at around @xmath104 m@xmath4yr@xmath5 ( with slightly higher values observed in cosmos which , we remind , includes on average brighter fir sources due to the lower sensitivity of the _ herschel _ observations both at 100 @xmath10 m and 160 @xmath10 m ) .
in addition , our data clearly show that star - forming galaxies are much smaller objects . indeed ,
while agn hosts have masses always @xmath105 m@xmath4 , star - forming galaxies can be found for stellar masses as low as @xmath106 m@xmath4 . on the other hand , while star - forming galaxies disappear beyond @xmath107 m@xmath4 , a non - negligible fraction of radio - selected agn can still be found beyond that value .
this result is better illustrated in figure [ fig : mass_hist_sfvsagn ] which indeed shows that , irrespective of the considered field , there is a trend for agn to be increasingly more numerous within the population of radio - selected sources as the stellar mass of the hosts increases ( cfr bottom panels of both plots ) .
so , are these differences between radio - selected agn and star - forming galaxies inbuilt in the process of their formation or do these two populations differentiate at a certain point of their evolution ?
the answer can be found in the plots presented in figures [ fig : mass_lir_gnandgs_z ] and [ fig : mass_lir_cosmos_z ] .
these show that , a part from a number of outliers which mostly represent objects set at very high ( z @xmath108 ) redshifts and mainly lie in the combined goodss+goodsn fields ( see also the discussion in 6 ) , the distributions of radio - selected agn and star - forming galaxies are almost indistinguishable one from the other down to z @xmath65 .
the two populations only differentiate in the most local universe , whereby agn are endowed with much larger masses and luminosities than star - forming galaxies .
these results are in agreement and extend the findings of the former sections which indicate that for redshifts above @xmath65 there is no difference between radio - selected agn and the sub - population of those which are also active at fir wavelengths . here
we have seen that , except for a few cases which mostly reside in the very early universe , at the same high redshifts there is also no difference between fir - active , radio - selected , agn and fir - active , radio - selected , star - forming galaxies . in all cases , differences appear only in the local universe : radio - active agn start progressively becoming fir - quiet , the more as the higher is their radio luminosity ( cfr 4 ) . at the same time , agn are found associated to much higher galaxy masses than star - forming galaxies .
interestingly enough , the few agn which still emit at fir wavelengths in the local universe exhibit higher fir luminosities than star - forming galaxies ( cfr left - hand plots of figures [ fig : mass_lir_gnandgs_z ] and [ fig : mass_lir_cosmos_z ] ) .
with the aim of providing an analysis as exhaustive as possible of the far infrared properties ( fir ) of radio - selected agn , we have considered three very well studied fields such as the lockman hole , the goods - n and the goods - s .
these fields are indeed provided with the deepest radio and fir observations available up - to - date and have also been observed at a large number of other wavelengths , allowing the determination of spectroscopic and photometric redshifts for the overwhelming majority of the sources residing there . + for the primary radio selection , we used data from ibar et al .
( 2009 ) , miller et al .
( 2013 ) and morrison et al .
( 2010 ) . in all cases , these reach depths of @xmath109jy at 1.4 ghz . in order to distinguish between radio emission due to star - formation ongoing in the host galaxy and radio emission of agn origin , we used the method already presented in magliocchetti et al .
( 2014 ) which is based on the radio luminosity of the sources .
this marks as an agn all those objects with 1.4 ghz luminosities brighter than a chosen threshold , which roughly corresponds to the break in the radio luminosity function of star - forming galaxies and which positively evolves with look - back time .
the number of radio - emitting agn found on the three considered fields with the above method is 92 : 45 in the lockman hole , 15 in the goods - s and 32 in the goods - n . by then relying on the very deep observations provided by the _
herschel _ mission at 100 @xmath10 m and 160 @xmath10 m , we used the pep ( lutz et al .
2011 ) and goods-_herschel _ ( elbaz et al . 2011 ) data to look for fir counterparts to radio - selected agn .
thanks to the extreme depth of fir observations , we found that the large majority of radio - emitting agn are indeed fir emitters .
the fraction of sources with ongoing fir activity reaches values as high as @xmath60% in the case of goods - n and @xmath2% in the lockman hole , where fir observations are shallower .
we found that possibly except for the very local , @xmath70 , universe the chances for fir emission in a radio - active agn do not depend on the redshift of the sources , implying that fir activity is a common event throughout the whole life - time evolution of these objects .
as shown in magliocchetti et al .
( 2014 ) , we confirm that fir emission in radio - active agn is entirely due to star - forming activity within the host galaxy .
the bolometric luminosities of these sources are extremely high : the peak of their distribution is observed at around l@xmath110 l@xmath4 , with agn reaching l@xmath52 values as high as 10@xmath111 l@xmath4 . converted into star formation rates ( sfr )
, this implies sources with average sfr @xmath112 m@xmath4yr@xmath5 which can even reach values up to @xmath113 m@xmath4yr@xmath5 . recalling that the large majority of radio - active agn are also found to be fir emitters
, we can then draw the conclusion that most of the galaxies which host a radio - active agn are also the hosts of extremely intense star - forming activity .
this finding confirms and extends to the radio - active case all recent results obtained on the very high star - forming activity observed in the hosts of agn selected in different ways and at various wavelengths ( e.g. hatziminaoglou et al .
2010 ; santini et al . 2012 ; rosario et al .
2013 just to mention a few ) .
perhaps more importantly , the data show that neither the spectral slope of the radio emission , nor the radio - luminosity of the sources under examination vary in the presence of fir emission .
this is true for the spectral slope at all redshifts , and suggests that the central engine responsible for the agn phenomenon is oblivious , at least at radio wavelengths , to the presence of ongoing star formation within the same galaxy . in other words ,
fir activity does not seem to influence radio agn activity .
this is also true for radio luminosities , at least for redshifts larger than @xmath54 : indeed we find that , irrespective of radio luminosity , in the more distant universe virtually all radio - emitting agn cohabit with a star - forming event . once again
, this suggests that radio luminosity , and therefore agn activity at radio wavelengths , is not affected by the nearby star - forming processes .
the situation varies in the more local universe , as there is a deficit of fir emitters amongst bright radio agn .
this is in accordance with the general view that powerful radio - loud agn reside in old galaxies with little or no ongoing star - forming activity .
our results then show that this lack of star - forming activity within the hosts of radio - active agn is only true in the nearby universe , and possibly implies that at some late stage of their evolution , and only for weak enough fir emission , powerful enough agn are capable to swipe off the gas in the surrounding star - forming regions and consequently switch off the on - site star - formation .
note that our results are in excellent agreement with those of best et al .
( 2014 ) , which also show a clear positive evolution of radio - loud agn which are in the radiative - mode ( i.e. associated to cold gas and therefore to star - formation ) with respect to those which are in the so - called jet - mode ( radiatively inefficient and generally associated to passive galaxies ) between @xmath114 and @xmath115 .
another possible interpretation of the data is provided by recent studies of the mass function of galaxies which show that there is a transition at around @xmath54 between a mass function dominated by late - type galaxies to one dominated by early - type ones ( e.g. pozzetti et al .
2010 ; ilbert et al .
the radio - agn behavior might then only mirror this more general trend derived for the whole galaxy population .
another relevant result of our analysis is that also the general properties of the galaxies host of radio - selected agn ) , such as their stellar mass or their age , do not vary in the presence of ongoing fir activity .
this means that the hosts of radio - active agn all belong to the same population , irrespective of whether or not there is ongoing star - forming activity , i.e. that star - formation is not the dominant ingredient in shaping a galaxy which also hosts a radio - active agn .
we stress once again that the above results rely on an extremely high success rate of radio - to - fir associations and therefore that the conclusions we can draw with the present work refer to the majority of radio - active agn . for this class of sources no effect of star - forming origin
is observed on agn activity .
all the considered properties both of the host galaxy and of the agn itself remain unaltered in the presence of fir emission , showing that star - formation does not influence ( at least radio ) agn activity or shape the physical properties of the hosting galaxies . on the other hand , most of such agn hosts ( virtually the totality for redshifts @xmath6 )
are also found to be the sites of even extremely intense star - forming activity .
this indicates that also star formation activity is not affected by the presence of an agn , in all cases except for radio - luminous sources in the local , @xmath11 , universe .
we can then conclude that , except for the case of relatively local sources and high radio luminosities , no interaction is observed between the processes of agn emission and star - formation , as they seem to proceed independently of each other .
note that no clear evidence for the dominance of agn - feedback in shaping galaxy evolution has also been found by pozzi et al .
( 2015 ) , who use a backward approach capable to reproduce the galaxy k - band and the far - ir luminosity functions across the redshift range @xmath116 .
lastly , we have shown that , except for a few cases of sources mainly residing at very high ( z @xmath108 ) redshifts , galaxies hosting radio - selected star - forming events and those hosting radio - selected agn are indistinguishable from each other both in terms of mass and ir luminosity ( or star - forming rate ) at all redshifts larger than @xmath65 .
differences only start appearing in the local universe , where radio - active agn start progressively becoming fir - quiet , the more as the higher is their radio luminosity , and at the same time are found associated to much larger galaxies than those which host a star - forming event . why differences in the various radio - selected populations considered here only appear in the local universe
is still unclear and will be investigated in a forthcoming paper .
+ + * acknowledgements * we wish to thank the referee for his / her careful reading of the manuscript and constructive suggestions which helped shaping and improving the paper .
pacs has been developed by a consortium of institutes led by mpe ( germany ) and including uvie ( austria ) ; ku leuven , csl , imec ( belgium ) ; cea , lam ( france ) ; mpia ( germany ) ; inaf- ifsi / oaa / oap / oat , lens , sissa ( italy ) ; iac ( spain ) .
this development has been supported by the funding agencies bmvit ( austria ) , esa - prodex ( belgium ) , cea / cnes ( france ) , dlr ( germany ) , asi / inaf ( italy ) , and cicyt / mcyt ( spain ) .
alexander et al .
2008 , apj , 135 , 1968 appleton p.n .
et al . , 2004 , apjs , 154 , 147 berta s. et al . , 2011 , a&a , 532 , 49 best p. n. , ker l. m. , simpson c. , rigby e. e. , sabater j. , 2015 , arxiv:14090263 bondi m. et al .
2003 , a&a , 403 , 857 bondi m. , ciliegi p. , schinnerer e. , smolcic v. , jahke k. , carilli c. , zamorani g. , 2008 , apj , 681 , 1135 boyle b.j . , cornwell t.j . , middleberg e. , norris r.p . ,
appleton p.n . ,
smail i. , 2007 , mnras , 385 , 1143 brinhcmann j. , charlot s. , white s.d.m .
tremonti c. , kauffmann g. , heckman t. , brinkman j. , 2004 , mnras , 351 , 1151 brunner h. et al .
2008 , a&a , 479 , 283 .
bruzual g. , charlot s. , 2003 , mnras , 344 , 1000 carilli c.l . , furlanetto s. , briggs f. , jarvis m. , rawlings s. , falcke h. , 2004 , new astronomy reviews , elsevier , vol .
48 , issue 11 - 12 , 1029 condon j.j . ,
condon m.a . , gisler g. , pushell j. , 1982 , apj , 252 , 102 condon jj . , 1992 ,
ara&a , 30 , 575 de breuck c. et al . , 2010 , apj , 725 , 36 del moro a. et al . , 2013 ,
a&a , 549 , a59 elbaz d. et al . , 2011 ,
a&a , 533 , a119 fanaroff b.l .
, riley j.m . , 1974 , mnras , 167 , 31 fotopoulou s. et al . , 2012 , apjs , 198 , 1 garn t. , green d.a . , riley j.m .
alezander p. , 2009 ,
mnras , 397 , 1101 griffin m.j .
et al . , 2010 , a&a , 518 , l3 gruppioni c. et al .
2013 , mnras , 432 , 23 hatziminaoglou e. , et al .
2010 , a&a , 518l , 33 ibar e. , ivison r.j .
, biggs a.d . ,
, best p.n . , green d.a .
, 2009 , mnras , 397 , 281 ibar e. , ivison r.j .
, best p.n . , coppin k. , pope a. , smail i. , dunlop j.s . , 2010 , mnras , 401 , l53 ilbert o. , 2013 , a&a , 556 , 55 johnston s. et al . , 2007 , pasa , 24 , 174 jonas j.l . , 2009 , ieee proc
97 , 1522 kennicut r.c .
jr , 1998 , ara&a , 36 , 189 leipski c. , antonucci r. , ogle p. , whysong d. , 2009 , 701 , 891 lutz d. et al .
2011 , a&a , 532 , 90 lutz d. 2014 , ara&a , 52 , 373 magliocchetti m. , maddox s. j. , wall j. v. , benn c. r. , cotter g. , 2000 , mnras , 318 , 1047 magliocchetti m. et al .
( the 2dfgrs team ) 2002 , mnras , 333 , 100 magliocchetti m. et al .
( the 2dfgrs team ) 2004 , mnras , 350 , 1485 magliocchetti m. , andreani p. , zwaan m.a . , 2008 , mnras , 383 , 479 magliocchetti m. et al . , 2014 , mnras , 442 , 682 magnelli b. et al . 2013 , a&a , 553 , a132 .
et al . , 2011 , apj , 731 , 79 mauch t. , sadler e.m . , 2007 , mnras , 375 , 931 mcalpine k. , jarvis m.j . , bonfield d.g .
, 2013 , mnras , 436 , 1084 merloni a. , rudnik g. , di matteo t. , 2004 , mnras , 354 , l37 metcalf r.b . ,
magliocchetti m. , 2006 , mnras , 365 , 101 miller n.a .
et al . , 2013 ,
apjs , 205 , 13 morrison g.e . ,
owen f.n . , dickinson m. , ivison r.j . , ibar e. , 2010 , apjs , 188 , 178 norris r.p .
et al . , 2011 ,
apj , 736 , 55 peng y.j .
et al . , 2010 ,
apj , 721 , 193 pilbratt g. et al . , 2010 , a&a , 518 , l1 poglitsch a. et al . , 2010 , a&a , 518 , l2 pozzetti l
.. et al . , 2010 , a&a , 517 , 11 pozzi f. et al . , 2015 ,
apj , 803 , 35 prandoni i. , gregorini l. , parma p. , de ruiter h.r . , vettolani g. , zanichelli a. , wieringa m.h . , ekers r.d . , 2001 , a&a , 369 , 787 rodighiero g. et al . 2011 , apj , 739 , l40 randall k.e . ,
hopkins a.m. , norris r.p . , zinn p .- c .
, middleberg e. , mao m.y .
, sharp r.g . , 2012 , mnras , 421
, 1644 rosario d. et al . , 2012 ,
a&a , 545 , 45 rosario d. et al .
, 2013 , a&a , 560 , 72 salpeter e. e. , 1955 , apj , 121 , 161 santini p. et al . , 2009 , a&a , 504 , 751 santini p. et al . , 2012 ,
a&a , 540 , 109 schinnerer e. , et al . , 2004 ,
aj , 128 , 1974 schinnerer e. et al .. , 2007 , apjs , 172 , 46 schinnerer e. et al .. , 2010 , apjs , 188 , 384 seymour n. et al . , 2011 , mnras , 413 , 1777 simpson c. et al . 2012 , mnras , 421 , 3060 | in order to investigate the fir properties of radio - active agn , we have considered three different fields where both radio and fir observations are the deepest to - date : goods - south , goods - north and the lockman hole . out of a total of 92 radio - selected agn , @xmath0 are found to have a counterpart in _ herschel _ maps .
the percentage is maximum in the goods - north ( @xmath1% ) and minimum ( @xmath2% ) in the lockman hole , where fir observations are shallower .
our study shows that in all cases fir emission is associated to star - forming activity within the host galaxy .
such an activity can even be extremely intense , with star - forming rates as high as @xmath3 m@xmath4yr@xmath5 .
agn activity does not inhibit star formation in the host galaxy , just as on - site star - formation does not seem to affect agn properties , at least those detected at radio wavelengths and for @xmath6 . given the very high rate of fir detections , we stress that this refers to the majority of the sample : most radio - active agn are associated with intense episodes of star - formation .
however , the two processes proceed independently within the same galaxy , at all redshifts but in the local universe , where powerful enough radio activity reaches the necessary strength to switch off the on - site star formation .
our data also show that for z @xmath7 the hosts of radio - selected star - forming galaxies and agn are indistinguishable from each other both in terms of mass and ir luminosity distributions .
the two populations only differentiate in the very local universe , whereby the few agn which are still fir - active are found in galaxies with much higher masses and luminosities .
galaxies : evolution - infrared : galaxies - galaxies : starburst - galaxies : active - radio continuum : galaxies - methods : observational |
the planck satellite is designed to measure temperature fluctuations in the cosmic microwave background ( cmb ) with a precision of @xmath1 , and angular resolution of about 5 arcminutes .
the payload consists of a 1.5 - 2.0 m gregorian telescope which feeds two instruments : the high frequency instrument ( hfi ) with 56 bolometer arrays operated at @xmath2 and frequencies of @xmath3ghz and the low frequency instrument ( lfi ) with 56 tuned radio receivers arrays operated at @xmath4 ( @xmath5 ) and frequencies of @xmath6 ghz ( see http://astro.estec.esa.nl/sa-general/projects/planck/ for more information ) .
data on board planck consist of @xmath7 differential temperature measurements , spanning a range of values we shall call @xmath8 . preliminary studies and telemetry allocation indicate the need for compressing these data by a ratio of @xmath9 . here
we will consider under what conditions it might be possible to achieve such a large compression factor .
a discretized data set can be represented by a number of bits , @xmath10 , which for linear analogue - to - digital converters ( adc ) is typically given by the maximum range @xmath11 : @xmath12 . if we express the joint probability for a set of n measurements as @xmath13 , we have that the shannon entropy per component of the data set is : h - 1 _ i_1 , , i_n p_i_1 , , i_n _ 2(p_i_1 , , i_n ) .
[ h ] shannon s theorem states that @xmath14 is a lower bound to the average length of the code units .
we will therefore define the theoretical ( optimal ) compression rate as c_r , opt [ cr ] for a uniform distribution of @xmath7 measurements we have @xmath15 and @xmath16 , which equals the number of bits per data .
thus : it is not possible to compress a ( uniformly ) random distribution of measurements .
gaztaaga ( 1998 ) , have argued that a _ well calibrated signal _ will be dominated by thermal ( white ) noise in the instrument : @xmath17 and therefore suggested that the digital resolution @xmath18 only needs to be as small as the instrumental rms white noise : @xmath19 .
the nominal @xmath20 pixel sensitivity will only be achieved after averaging ( on earth ) .
this yields compression rates of @xmath21 . on the other hand maris ( 1999 )
have used the same formalism as gaztanaga but fixing the final dynamical range @xmath8 to some fiducial values , so that the digital resolution is then given by @xmath22 , independently of @xmath23 . again , assuming a well calibrated signal dominated by thermal ( white ) noise , this approach yields smaller compression rates of @xmath24 , as it is obvious from the fact that the digital resolution is larger ( @xmath18 is smaller ) . in both cases ,
the effect of the cmb signal ( eg dipole ) and other sources ( such as the galaxy ) have been ignored .
several questions arise from these studies . what is the optimal value of @xmath18 and what are the penalties ( distortions ) involved when using large values of @xmath18 ? moreover , can the data gathered by the on board instruments be really modeled as a white noise signal ? in other words , are the departures from gaussianity ( due to the galactic , foregrounds , dipole and cmb signals ) important ?
this latter question is closely related to the way data will be processed ( and calibrated ) on board , for example : if and how the dipole is going to be used for calibration .
these issues together with the final instrument specifications seem to play an important role on the final range of values @xmath8 and , therefore , the possible compression rates .
this is somehow unfortunate as compression would then be related in a rather complicated way to the nature of the external signal and also to critical issues of the internal data processing issues .
here we shall present a simple way of coding the on board data that will solve the lossless compression problem in a much simpler way .
this will be done independently of the internal calibration or the nature of the external signal ( cmb or otherwise ) .
we will also address the issue of the digital distortion introduced ( the penalty ) as a function of the final compression ( the prize ) . in section
2 we give a summary of some critical issues related to the on - board data .
our coding and compression proposals are presented in 3 , while simulations are dicussed in 4 .
we end up with some concluding remarks .
parameters for the radiometers : a ) central frequency @xmath25 ( bandwidth is 20% ) ; b ) angular resolution ( beam fwhm ) ; c ) rms thermal noise expected at 6.9 ms ( 144.9 hz ) sampling ; d ) range of temperatures expected from the sky ( jupiter , dipole , s - z ) ; e ) number of detectors ( 2x horns ) ; f ) total data rate at 6.9 ms ( 2.5 arcmin ) . [ cols="<,<,<,<,<,<",options="header " , ] we can model the process of differencing as the subtraction of two gaussian random variables : @xmath26 and @xmath27 with variances @xmath28 and @xmath29 .
the probability density distribution for the difference random variable @xmath30 is also a gaussian distribution with a new variance @xmath31 : @xmath32 for a wide sense stationary process @xmath33 and @xmath34 .
one can obtain also in this way the expression for the entropy of the distribution , @xmath35 we want to take differences of data separated by @xmath36 minute , which corresponds to the same sky position .
bearing in mind that our model is a first order markov process @xmath37 will be equal to the correlation between pixels separated 1 min .
, that is @xmath38 ( recall that the correlation matrix for a wide sense stationary stochastic process is a symmetric toeplitz matrix and so it depends only on index differences ) .
thus the two - point correlation is : @xmath39 next , we are going to estimate this correlation for a power spectrum @xmath40 of the type of white noise plus @xmath41 ( i.e. @xmath40 in eq.[[pk ] ] ) . in practice
our spectrum will not run over the whole range but only over a limited interval @xmath42 .
the final result , for @xmath43 , is : @xmath44_{f = f_{min}}^{f = f_{max}}\ ] ] where @xmath45 we want to know if correlation due to @xmath41 noise is important in our data handling model , so let us calculate some specific examples . in our model @xmath46
is given by the inverse of the sampling rate and @xmath47 is the inverse of two hours ( if calibration occurs at every pointing ) or the inverse of the mission s time ( about 1 year ) . in the table [ table :
rho ] we have computed the magnitude of @xmath37 for our model and for two different values of the calibration time , i.e. @xmath48 .
we can see in the table how small the values of @xmath37 are compared to unity .
given the precision needed for the entropy and compression factors , such contributions of @xmath37 to @xmath49 are neglegible . | we present a simple way of coding and compressing the data on board the planck instruments ( hfi and lfi ) to address the problem of the on board data reduction .
this is a critical issue in the planck mission . the total information that can be downloaded to earth is severely limited by the telemetry allocation .
this limitation could reduce the amount of diagnostics sent on the stability of the radiometers and , as a consequence , curb the final sensitivity of the cmb anisotropy maps . our proposal to address this problem consists in taking differences of consecutive circles at a given sky pointing .
to a good approximation , these differences are independent of the external signal , and are dominated by thermal ( white ) instrumental noise . using simulations and analytical predictions
we show that high compression rates , @xmath0 , can be obtained with minor or zero loss of cmb sensitivity .
possible effects of digital distortion are also analized .
the proposed scheme allows for flexibility to optimize the relation with other critical aspects of the mission .
thus , this study constitutes an important step towards a more realistic modeling of the final sensitivity of the cmb temperature anisotropy maps .
# 1 # 1([#1 ] ) # 1 # 1 # 1 # 1 # 1 # 1 # 1 # 1 # 1to 0pt#1 |
topological solid states of matter @xcite have the property that their bulk states are gapped , while states at the boundaries are gapless and protected by some discrete quantum symmetry .
the topological aspect emerges when considering the transport properties of the boundary states , where the transport current happens to also be a topological current .
the most well established topological solid states of matter are topological insulators ( tis ) , which are gapped in the bulk and have helical ( or chiral ) gapless states at the boundaries which are protected by time - reversal symmetry .
helical here means that the electronic spin is locked to momentum due to strong spin - orbit coupling .
thus , the boundary states have an helicity determined by the eigenvalues of @xmath1 at each boundary .
topological insulators have been predicted to exist @xcite and confirmed experimentally in subsequent papers @xcite .
although many of the materials investigated experimentally are not perfect insulators in the bulk , the observed boundary helical states are robust features of these materials .
there exists another predicted topological solid state of matter , namely topological superconductors ( tscs ) @xcite , where the experimental situation is less clear .
tscs follow a symmetry classification scheme closely related to tis , as far as hamiltonians of the bogoliubov - de gennes type are concerned.@xcite just like tis , tscs have gapped states in the bulk and symmetry - protected gapless states at the boundaries .
unlike tis , in tscs the @xmath2 symmetry is broken , either spontaneously or by proximity effect .
the gapless boundary states are majorana fermions , which are fermionic particles that are their own anti - particles . in order to support such states at the boundaries ,
the topological superconductivity must feature a @xmath3-wave type of gap .
particle - hole is the underlying symmetry protecting the boundary majorana states . in one dimension ,
a paradigmatic simple model for topological superconductivity has been proposed by kitaev @xcite where the majorana zero - energy states live at the ends of a quantum wire .
an experimental way of realizing a superconducting state in a quantum wire is by proximity effect . in this case
a semiconducting wire with strong spin - orbit coupling is deposited on the surface of an @xmath4-wave superconductor in the presence of an external perpendicular magnetic field .
then by proximity effect @xmath3-wave like superconductivity is induced on the wire for a certain range of parameters.@xcite there are some experimental signatures of majorana modes in indium antimonide nanowires in contact with normal and superconducting electrodes.@xcite more recently , strong support for majorana boundary zero modes have been reported in an experiment with a ferromagnetic chain of iron fabricated on the surface of lead.@xcite three - dimensional ( 3d ) tscs have also been discussed theoretically , in particular focusing on vortex physics @xcite and possible topological phase transitions.@xcite a distinctive feature of both three - dimensional tis and tscs with respect to their non - topological counterparts is the topological magnetoelectric response induced by a mechanism similar to the so - called chiral anomaly.@xcite when fermions in topological materials interact with the electromagnetic field , a berry phase mixing electric and magnetic fields is induced.@xcite in tis this occurs due to strong spin - orbit coupling that locks spin to momentum .
the resulting berry phases combine in the form of a so - called axion term , which is a magnetoelectric term @xmath5 with a periodic field , @xmath6 , appearing as a coefficient.@xcite this coefficient corresponds to a topological invariant implied by the chern number of the band structure .
if tr symmetry is preserved , there are only two possible values for @xmath6 in a 3d insulator , namely , @xmath7 and @xmath8 , the former value corresponding to a topologically trivial insulator.@xcite in the case of tscs a topological magnetoelectric contribution also arises , but now @xmath6 corresponds to the phase difference between order parameters of opposite chirality.@xcite recently , axionic superconductivity has been also discussed in the context of doped narrow - gap semiconductors.@xcite the 3d tsc constructed by qi _
@xcite should actually correspond to an interacting topological state of matter , going beyond the the symmetry classification of topological non - interacting theories for insulators and superconductors.@xcite according to the standard symmetry classification and an argument following from the gravitational anomaly,@xcite 3d tscs having tr invariance are classified by an integer topological invariant , belonging to the diii class in the free fermion classification table.@xcite thus , unlike tr invariant tis , the electromagnetic axionic response of 3d tscs would involve an axion field having values @xmath7 or @xmath8 , mod @xmath9 . however , as pointed out in ref .
, the axion in tscs is a dynamical phase variable associated with the superconducting order parameter , or higgs field in the language of quantum field theory .
furthermore , the meissner effect gives a mass to the photon via the higgs mechanism .
this is an important difference with respect to the axion electrodynamics of tis , where no u(1 ) symmetry is broken and the gauge field remains gapless .
furthermore , nontrivial chern numbers are associated with the different phases . for the simpler case featuring two weyl fermions with opposite helicity
, we have opposite chern numbers , leading to a topological invariant given by the sign of the gap amplitude.@xcite thus , the arguments in ref . seems to point out to a @xmath10 classification .
however , since the relevant symmetry in the problem is @xmath11 , it has been shown recently @xcite that in topological superconductors with tr symmetry the @xmath12 classification is reduced to @xmath13 . in this paper
, we investigate the higgs mechanism and anomalous hall effect of three - dimensional tscs within the model introduced recently in ref . .
in the simplest case , the model features two superconducting order parameters associated with left and right fermion chiralities interacting with the electromagnetic field , and a topological magnetoelectric term .
we will show that quantum fluctuations in such a sc imply an interacting topologically non - trivial phase that can not be continuously deformed into the interacting topologically trivial one .
our claim is substantiated by a renormalization group analysis .
this result does not follow from the classical lagrangian of the system derived earlier in ref .
, and is a purely quantum effect involving the interaction between photons and higgs fields . we show that due to this behavior , the type i regime of the tsc features a first - order phase transition in the bulk and a second - order phase transition on the surface .
this distinguishes a tsc from a topologically trivial superconductor in the type i regime .
the latter would exhibit a first - order phase transition in the bulk as well as on the surface , provided quantum fluctuations are accounted for @xcite ( this point will be discussed in detail in section iii ) .
another consequence of the topological magnetoelectric term is the occurrence of an anomalous hall effect when an electric field is applied parallel to the surface of a tsc . due to the magnetoelectric effect
, a transverse current is generated from a lorentz - like force involving the relative superfluid velocity @xmath14 and the applied electric field .
the generated transverse current is negative , a situation reminiscent of the anomalous hall effect in superconductors @xcite , and observed high-@xmath0 cuprate superconductors.@xcite however , in the latter case the anomalous hall effect occurs due to vortex motion induced by the faraday law , and is typically a very small effect .
furthermore , in this case the lorentz force acts directly on the vortex core , and therefore on the normal components of the superconductor .
for this reason , it automatically leads to dissipation . in the case of three - dimensional tscs ,
on the other hand , the anomalous hall effect occurs even in absence of vortices and is induced solely by an external electric field via the magnetoelectric effect .
dissipationless anomalous hall current is generated on the surface_. the plan of the paper is as follows .
in section ii we discuss how the higgs mechanism works in a topological superconductor .
we will show that after the phases are integrated out , interactions between the photons automatically arise due to the axion term .
this is in contrast to the ordinary higg mechanism , where the phases can be trivially integrated out by a gauge transformation . in section iii
we discuss the role of quantum fluctuations and derive the effective potential on the surface and the renormalization group ( rg ) equations .
we show that the rg equations of the topological superconductor can not be connected to the ones of a topologically trivial superconductor .
this is shown to occur as a direct consequence of the axion term . finally , in section iv we discuss the dissipationless variant of the anomalous hall effect using the london limit of topological superconductors .
the effective lagrangian for a three - dimensional tsc featuring two fermi surfaces is given by @xcite , @xmath15\nonumber\\ & -&\frac{u}{2}(|\phi_l|^2+|\phi_r|^2)^2 + 2j(\phi_l^*\phi_r+\phi_r^*\phi_l),\end{aligned}\ ] ] where @xmath16 is the charge of the condensate .
the greek indices run from 0 to 3 and @xmath17 , with @xmath18 .
the lagrangian ( [ eq : l - higgs ] ) corresponds to an abelian higgs model with a two - component scalar field @xmath19 , where @xmath20 denote `` right '' and `` left '' chiralities of the two components of the scalar condensate matter field , minimally coupled to the gauge - field @xmath21 .
in contrast to the standard higgs model , eq .
( [ eq : l - higgs ] ) features a so - called axion term @xcite , which is the first term appearing in the lagrangian above .
the term is topological in nature and contains a scalar field ( the axion ) @xmath22 , where @xmath23 and @xmath24 are the phases of @xmath25 and @xmath26 , respectively , i.e. @xmath27 . in terms of electric and magnetic field components ,
the toplogical term reads @xmath28 , which is precisely the magnetoelectric form mentioned in the introductory paragraphs .
a josephson coupling term @xmath29 accounts for the interference between the two superconducting order parameter fields .
this is a characteristic feature in superconductors with two or more components of the order parameter , and is absent only if prohibited by symmetries of the problem .
@xcite below , we show that the josephson coupling is generated by fluctuations , and therefore it is legitimate to include such a term from the very beginning in the lagrangian .
furthermore , the josephson coupling is important for tuning between topologically trivial and nontrivial phases .
in fact , a simple mean - field analysis shows that for @xmath30 the josephson coupling implies @xmath7 , yielding a topologically trivial superconductor . for @xmath31 ,
on the other hand , @xmath6 is locked to @xmath8 , thus leading to a topologically non - trivial superconducting ground state .
since @xmath6 is periodic , @xmath32 corresponds to a situation where the time - reversal ( tr ) symmetry is preserved @xcite .
thus , at the mean - field level , @xmath33 separates a topologically trivial ground state from a non - trivial one .
thus , varying @xmath34 from positive to negative values induces a topological quantum phase transition
. in the @xmath2 higgs mechanism the phases disappear from the spectrum due to spontaneous breaking of the local @xmath2 symmetry , being transmuted into the longitudinal mode for the photon , which becomes gapped .
thus , only amplitude modes remain in the spectrum of scalar particles .
the higgs mechanism is equivalent to integrating out the phase degrees of freedom , which in the case of the higgs model automatically leads to a massive gauge particle .
this point of view of integrating out the phases is particularly appealing in the case where a josephson coupling is present .
however , additional non - linearities arise in the presence of the axion term . to see this ,
let us first consider the higgs mechanism in eq .
( [ eq : l - higgs ] ) for the case where the axion term is absent . in this case
we can simply write @xmath35 , @xmath36 and make the shift , @xmath37 which yields , @xmath38 note that in absence of the axion term and for the particular situation of a single scalar field component , i.e. , @xmath39 and @xmath40 , the above effective lagrangian trivially reduces to the usual lagrangian for the higgs model in the unitary gauge . in this particular case
the lagrangian is independent of the phase . for the case relevant to us here ,
where two scalar fields are present , there is a term @xmath41 remaining .
thus , in the absence of josephson coupling there is still a massless ( goldstone ) mode present in the spectrum .
this occurs because there are two higgs fields and one abelian gauge field .
thus , it is only possible to gauge away one phase degree of freedom .
the phase @xmath6 would not couple directly to the gauge field in the absence of the axion term . due to the axion term , integrating out the phases generate direct interactions between photons , even if the amplitudes are assumed to be uniform .
for the topological phase of the system occurring for @xmath31 , we have to integrate out the lowest order gaussian phase fluctuations around @xmath32 . in general
, this renders the induced photon - photon interaction being non - local .
if the amplitudes are uniform , we obtain the effective lagrangian , @xmath42 where @xmath43 with @xmath44 . in the long wavelength regime
the induced photon interaction is strongly screened by the axion .
hence , we have @xmath45 .
the resulting photon - photon interaction simplifies and we obtain@xcite @xmath46 where @xmath47 .
we now turn to a crucial aspect of the topological phase with respect to the surface states .
it turns out that in the presence of quantum fluctuations , the topological surface states can not be continuously deformed into topologically trivial ones in the long wavelength limit when crossing the critical point . to see , this let us assume that @xmath6 is uniform on the surfaces , with @xmath32 for the tr invariant case .
note that @xmath23 and @xmath24 are still allowed to fluctuate , with @xmath48 and @xmath49 .
since @xmath50 , each surface contains a chern - simons ( cs ) term.@xcite thus , assuming two surfaces perpendicular to the @xmath51-axis , we find that the imaginary time photon propagator at any surface is given by @xmath52,\ ] ] where @xmath53 , @xmath54 , @xmath55 , and the @xmath56 sign is chosen depending on which surface one is referring to . in eq .
( [ eq : photon - prop ] ) the transverse gauge has been fixed . at the phase transition to the normal state
where @xmath57 , the feynman diagram shown in fig .
[ fig : fish ] , which is associated with higgs scattering mediated by photons , behaves very differently at long wavelengths ( @xmath58 ) , depending on whether @xmath7 ( topologically trivial ) or @xmath59 ( topologically non - trivial ) .
namely , for all @xmath59 we have , @xmath60 whereas the result is divergent for @xmath7 .
thus , after other one - loop scattering amplitudes are taken into account to obtain the full four - higgs vertex , we see that at the critical point the topological field theory can not be continuously deformed into a topologically trivial one .
this statement holds trivially for topological bogoliubov - de gennes superconductors . here
, we have shown that it also holds in the presence of quantum fluctuations in an interacting system , beyond the bogoliubov - de gennes picture .
this occurs because the photon is topologically gapped , despite the vanishing of the meissner gap ( @xmath61 ) .
note that this result is due to the topological character of the axion term , and not due to a symmetry protection .
indeed , since the diagram of fig .
[ fig : fish ] vanishes for any @xmath59 , tr invariance is not required .
thus , in this context the topologically non - trivial phase simply corresponds to the case where the axion term is nonzero .
a continuous deformation to the topologically trivial phase can be done in the higgs phase , where there are no gapless modes . at the critical point
such a continuous deformation is not possible .
thus , quantum critical fluctuations in this system will govern topologically stable universal behavior in physical quantities .
this has important implications for critical exponents and amplitude ratios . indeed , as we will see , the vanishing of the diagram shown in fig .
[ fig : fish ] changes significantly the renormalization group ( rg ) @xmath62 function associated to the interaction vertex between scalar fields . )
the corresponding feynman diagram diverges . on the other hand , for the topologically non - trivial case ( @xmath59 ) , the same diagram vanishes .
this shows beyond the non - interacting regime that a topological superconductor can not be continuously deformed into a topologically non - trivial one in the long wavelength limit.,width=302 ] as is well known , the one - loop effective action is more easily obtained by integrating out the quadratic fluctuations of the scalar fields and gauge fields.@xcite we will assume that the magnitudes of both @xmath25 and @xmath26 have the same expectation value in the broken symmetry phase .
thus , we write @xmath63 and @xmath64 , where @xmath65 is uniform and @xmath66 represents the fluctuation around @xmath67 . the effective action is therefore written in the form , @xmath68,\ ] ] where @xmath69,\ ] ] with @xmath70 being the ( infinite ) volume of three - dimensional spacetime , and @xmath71 $ ] .
the matrices @xmath72 and @xmath73 are given in momentum space by , @xmath74,\ ] ] where @xmath75 is a @xmath76 identity matrix , while @xmath77 and @xmath78 are pauli matrices , and @xmath79 from eq .
( [ eq : m ] ) , we see that a correction to the josephson coupling has been generated by fluctuations .
it has the form , @xmath80 .
this term is generated even if @xmath33 , which means that including a josephson coupling in the lagrangian is a physically reasonable assumption .
this should be expected , since fluctuations will necessarily lead to an overlap between the two complex field components .
note that this result is valid in general for any two - component superconductor and is not restricted to the topological one being considered here .
integrating out the fluctuations in eq .
( [ eq : gauss - flucts ] ) , we obtain , @xmath81 where @xmath70 is the three - dimensional spacetime volume and @xmath82 is the effective potential given by , @xmath83.\ ] ] the tracelogs can be written in more explicit form , @xmath84 \nonumber\\ & + & \frac{1}{2}\sum_{\sigma=\pm}\int\frac{d^3p}{(2\pi)^3}\left\{\ln\left[p^2+m_\sigma^2(|\varphi|,\theta)\right ] \right.\nonumber\\ & + & \left.\ln\left[p^2+m_{1\sigma}^2(|\varphi|,\theta)\right]+\ln\left[p^2+m_{2\sigma}^2(|\varphi|,\theta)\right ] \right\},\end{aligned}\ ] ] where , @xmath85 @xmath86 @xmath87 from this mass spectrum underlying the effective potential , we recognize @xmath88 and @xmath89 as the would - be goldstone modes from the regime @xmath33 , corresponding to the absence of the josephson coupling .
note that the corresponding uncharged system in absence of josephson coupling features three goldstone modes .
this is to be expected , since the system in this case would be @xmath90 invariant . as usual , in the charged system there are only gapped modes , even in absence of josephson coupling , as required by the higgs mechanism .
the momentum integral in eq .
( [ eq : u ] ) can be easily evaluated using an ultraviolet cutoff @xmath91 , @xmath92 \nonumber\\ & -&\frac{1}{12\pi}\sum_{\sigma=\pm}\left[|m_\sigma(|\varphi|,\theta)|^3+|m_{1\sigma}(|\varphi|,\theta)|^3 \right.\nonumber\\ & + & \left .
|m_{2\sigma}(|\varphi|,\theta)|^3\right],\end{aligned}\ ] ] where @xmath93 is a field - independent ( vacuum ) contribution .
we have absorbed a term @xmath94 in @xmath95 , since this contribution just corresponds to tadpole diagrams .
note that this simple renormalization does not actually affect the other terms , since the error committed by doing this is of higher order than one - loop . from the effective potential above
we see that the diagram of fig .
[ fig : fish ] can only arise from the sum @xmath96 , with @xmath97 given in eq .
( [ eq : m+- ] ) .
an expansion in powers of @xmath98 for @xmath59 clearly shows that the diagram shown in the figure indeed vanishes when @xmath59 .
however , the contributions @xmath99 for @xmath100 become singular _ for all _ @xmath6 .
we emphasize that this is only true when the topological magnetoelectric term is nonzero . on the other hand , if @xmath7 , we obtain , @xmath101 meaning that the effective potential is non - analytic as a function of @xmath102 .
albeit simple , this is actually a non - perturbative result , since it can not be obtained as a power series in @xmath103 .
the absence of a power series involving terms @xmath99 reflects the divergence of the diagram shown in fig.[fig : fish ] when @xmath7 for vanishing external momenta . in this case a meaningful perturbative evaluation of the vertex function can only be done for nonzero external momenta , a fact related to the infrared divergence arising from a massless photon .
note that in contrast to the @xmath59 case , the expansion is singular in the infrared for all @xmath104 , rather than for @xmath100 . however , the singularities in the case of a topologically massive photon is not a problem , since they correspond to interactions that are irrelevant in an rg sense .
for instance , this type of infrared singularity in higher order vertices would also occur in a simple @xmath105 landau theory .
having @xmath59 turns the photon topologically massive without spoiling gauge invariance.@xcite thus , we can interpret the cubic contribution arising in the limit of vanishing @xmath6 as a consequence of resumming up all the one - loop infrared divergent diagrams containing only internal photon lines .
this leads to a non - analytic contribution to the effective potential . as is well - known , summing up these contributions in @xmath106 in the case of standard scalar electrodynamics leads to a logarithmic term @xmath107 in the effective potential,@xcite while the cubic term has been obtained in the context of ginzburg - landau superconductors for @xmath108 ( i.e. , scalar electrodynamics in @xmath109 and imaginary time).@xcite in both cases these one - loop photon contributions lead to a fluctuation - induced first - order phase transition
however , it was later shown that this result is valid only in the type i regime , while in the type ii regime a second - order phase transition in the so - called inverted 3d xy universality class arises.@xcite in view of the above discussion , it is of interest to investigate the character of the superconducting phase transition on the surface of a topological superconductor .
the theory features two ingredients that are not present in the previous analysis of fluctuation - induced first - order phase transitions by halperin _
et al_. @xcite these are the josephson coupling between the scalar field components , and the cs term .
the case with josephson coupling and in absence of cs term has been examined in the london limit ( i.e. , in the strong type ii regime ) in ref . .
for this case it has been shown by means of exact duality arguments that a two - component superconductor with josephson coupling exhibits a phase transition in the 3d xy universality class.@xcite on the other hand , when the josephson coupling is absent , it has been shown using renormalization group ( rg ) methods in ref .
that for a large enough cs coupling ( i.e. , @xmath110 in our notation ) , the first - order phase transition is turned into a second - order one . regarding the result in the london limit without cs term , we note that our analysis in this section is being done in the type i regime , since amplitude fluctuations play an important role in the calculations above . indeed , the theory with @xmath111 yields a fluctuation induced first - order phase transition .
it is a well - known fact that the cs term does not renormalize.@xcite this is a consequence of the topological nature of the cs term . indeed , since it is independent of the metric , it does not change under scale transformations .
thus , using this result and the invariance of the effective action under renormalization , we obtain , @xmath112 where @xmath113 is the renormalized gauge field with the corresponding wavefunction renormalization , @xmath114 , and @xmath115 is the renormalized topological mass . from the above equation it follows that @xmath116 .
since gauge invariance implies the renormalization @xmath117 , @xcite which is easily obtained from the vacuum polarization , it follows that @xmath6 is a renormalization group invariant , @xmath118 where @xmath119 is a logarithmic renormalization scale defined in terms of the renormalized mass , @xmath120 , yielding the inverse correlation length .
( [ eq : rg - inv ] ) is an important result , since it allows one to study the critical topological behavior with a vanishing renormalized josephson coupling @xmath121 , while still having @xmath59 , corresponding to the critical point of the topological phase transition . combining both cs terms stemming from the two surfaces , we obtain that the axion term in the bulk does not get renormalized either , so the result ( [ eq : rg - inv ] ) also holds in the bulk .
an interesting consequence of this analysis is that the rg flow of the bulk theory does not differ significantly from the one where the axion term is absent , which is just given by the well - known analysis of coleman and weinberg.@xcite in this case , the phase transition is known to be of first - order , irrespective of the superconducting regime being of type i or type ii .
however , this is not the case for the phase transition on the surfaces .
indeed , the vanishing of feynman graph of fig .
[ fig : fish ] when @xmath59 implies that a contribution @xmath122 is absent in the rg @xmath62 function of @xmath123 . here are @xmath124 and @xmath125 renormalized dimensionless couplings defined on the surface of a tsc .
the one - loop rg functions for a superconductor in 2 + 1 dimensions with a cs term , and @xmath126 complex order parameter fields , are obtained in a way similar as in ref .
, except that we use the propagator ( [ eq : photon - prop ] ) with @xmath61 in the feynman diagrams involving photon lines .
the result is , @xmath127 @xmath128\hat u+\frac{(n+4)}{8\pi}\hat u^2,\ ] ] where we now have , @xmath129 where @xmath130 are the chern numbers associated to the helicity of the @xmath126 fermi surfaces involved.@xcite note that we have not expanded @xmath131 in powers of @xmath132 , because @xmath6 can also be very large , so that the product @xmath133 is not necessarily small
. it is easily seen that the above rg equations have an infrared stable fixed point for all @xmath126 , in contrast to the analysis by halperin _
et al . _ for the non - topological superconductor , where infrared fixed points are only found for @xmath134.@xcite however , there is a stability condition involving @xmath6 that has to be fulfilled in the case of a tsc .
it is obtained by considering the critical correlation function of the superconducting order field components , @xmath135 where at one - loop , @xmath136 which implies the inequality @xmath137 .
the latter inequality is fulfilled provided , @xmath138 a quantum critical point is obtained for all values @xmath6 , showing that at least sixteen weyl fermions are necessary to have a quantum critical point
. values of @xmath6 violating the inequality @xmath137 correspond to a situation where a continuum limit can not be defined and is therefore unphysical .
thus , in order to have a physically meaningful phase transition , the inequality ( [ eq : ineq ] ) has to be satisfied .
we note that the lower bound for @xmath139 is larger than @xmath8 when @xmath140 .
therefore , the tr symmetric value @xmath32 obtained at the mean - field level does not produce a second - order phase transition when quantum fluctuations are accounted for .
the one - loop rg predicts that a second - order phase transition occurs for @xmath32 only if @xmath141 .
it is tempting to relate the critical value of @xmath126 to the @xmath13 classification.@xcite however , at this stage it would be too speculative , since higher order results may affect the values of @xmath126 for which critical points obey the inequality @xmath137 . as a final remark on the rg analysis ,
let us comment on another possible renormalization scheme allowing to continuously connect the cases @xmath7 and @xmath59 .
the infrared divergences stemming from the photon propagator for @xmath7 requires defining the renormalized four - point vertex at external nonzero momenta .
this yields a renormalization scale @xmath142 that replaces @xmath120 in the rg flow . since the diagram of fig .
[ fig : fish ] does not vanish for nonzero external momenta , it turns out that a @xmath143 term would be generated in the rg @xmath62 function of @xmath123 , even when @xmath59 .
this @xmath143 term would have a @xmath6-dependent coefficient allowing to smoothly connect the result to the known rg equations of a topologically trivial superconductor in the limit @xmath144 .
we would find once more that a second - order phase transition occurs for large enough @xmath6 and a stability criterion would follow from the inequality @xmath137.@xcite the latter inequality would yield in this case values of @xmath6 leading to a _ negative _ coefficient of the @xmath143 term , with values of @xmath6 yielding a positive coefficient of @xmath143 violating the condition @xmath137 .
this implies that the rg @xmath62 function of @xmath123 can actually not be continuously connected to the @xmath7 regime , since to this end it would be necessary to enter a regime where the continuum limit is not even defined .
there are two important consequences of quantum fluctuations as unveiled by the analysis in this section .
first , we note that it is not possible to reach the topologically trivial phase from the topologically nontrivial one within the rg . simply taking the limit @xmath144
does not recover the rg flow of topologically trivial superconductors , while this limit can be realized classically .
second , and most importantly , the phase transition in the bulk is always a first - order one , while a second - order phase transition is possible on the surface .
this is not the case for the topologically trivial superconductor , where a second - order phase transition on the surface is only obtained for sufficiently large @xmath126 .
we next turn to the meissner effect aspects of a tsc , which as we will now show , implies an anomalous hall effect even in the absence of vortices .
this is more conveniently done by rewriting the lagrangian in a london limit exhibiting explicitly electric and magnetic fields , i.e. , @xmath145 from the effective lagrangian we obtain that the electric displacement and magnetic fields are given respectively by @xmath146 and @xmath147 , while the superconducting current is given by , @xmath148.\ ] ] from eq .
( [ eq : current ] ) we obtain the usual london equation in absence of vortices , @xmath149 , where @xmath150 is the square of the penetration depth .
thus , the maxwell equation in the presence of the axion field , @xmath151 yields the equation determining the london electrodynamics of the tsc in the form , @xmath152.\end{aligned}\ ] ] for the axion field we obtain the equation of motion , @xmath153 in the low frequency regime and in absence of vortices , the london equation ( [ eq : london ] ) simplifies to , @xmath154,\ ] ] while the current satisfies , @xmath155 the london equation for the electric field is unaffected by the axion term , retaining its traditional form , @xmath156 .
this result is closely related to the fact that the electromagnetic energy density does not contain a magnetolectric term .
we now consider a solution with a simple geometry , namely , a semi - infinite tsc ( @xmath157 ) with a surface at @xmath158 at an external electric field @xmath159 parallel to the surface .
we obtain , @xmath160 where @xmath161 , and @xmath162,\ ] ] @xmath163 the solution for eq .
( [ eq : jx ] ) in terms of the axion is , @xmath164,\ ] ] where @xmath165 . since @xmath166 , we obtain the following relation , @xmath167,\ ] ] which implies @xmath168 .
thus , we find that the usual boundary condition of the london theory , @xmath169 , is obviously fulfilled by the solution ( [ eq : sol - jy ] ) in the presence of the axion field
. however , the maxwell equation ( [ eq : maxwell ] ) in the static regime implies a boundary condition that deviates from the standard one in the london theory of non - topological superconductors , @xmath170 from eqs .
( [ eq : sol - jy ] ) and ( [ eq : relation ] ) we see that an approximate solution can be obtained by considering terms proportional to @xmath171 as being of higher order , which amounts to approximating eq .
( [ eq : sg-1 ] ) as being homogeneous . in this case
we can use the domain wall solution @xmath172 $ ] in eq .
( [ eq : sol - jy ] ) , which yields @xmath173 explicitly .
the explicit solution for @xmath173 with @xmath174 in terms of hypergeometric and lerch transcedents , is not very illuminating .
instead , we plot it in fig . [
fig : anom - hall ] for four different values of the ratio @xmath175 .
it has a negative sign , just like in the case of the anomalous hall effect in high-@xmath0 superconductors @xcite .
as emphasized in the introductory paragraphs , the anomalous hall effect in non - topological superconductors has a quite different origin from the one discussed here .
_ in three - dimensional tscs the anomalous hall current arises independently of vortex motion and is associated with a dissipationless current . _
for @xmath176 .
, width=340 ] for @xmath177 ( blue curve in fig .
[ fig : anom - hall ] ) , the expression for @xmath173 does not involve special functions , reading , @xmath178\right\}.\end{aligned}\ ] ] in order to make connection with the quantization of hall conductivity in the normal state , it is instructive to integrate the current density over @xmath179 to obtain the surface current density , @xmath180 we obtain , @xmath181,\end{aligned}\ ] ] where @xmath182 is the digamma function .
the normal state corresponds to @xmath183 , which yields @xmath184 , i. e. , the half - quantum of the quantized hall conductivity .
in conclusion , we have shown that due to quantum electromagnetic fluctuations , the higgs mechanism in three - dimensional tscs implies a robust topological state of matter , since its rg flow can not be continuously deformed into the rg flow of a topologically trivial one .
this is an example of a topological state that is protected due to the coupling of phase and electromagnetic fluctuations via the axion term , with tr symmetry not being required .
in fact , tr can be spontaneously broken by quantum fluctuations . in this context , we have also shown that a second - order quantum phase transition happens on the surface of a tsc , while its bulk undergoes a first - order phase transition . without the axion term a first - order phase transition would happen both in the bulk and on the surface ,
provided the superconductor is in the type i regime .
another aspect of the higgs mechanism we have studied is the influence of the axion term in the meissner effect .
we have found that the gradient of the axion field on the surface induces a transverse supercurrent . in the low frequency limit
this implies a london regime leading to the generation of an anomalous hall current with a negative sign .
this anomalous hall current is dissipationless and is the consequence of a lorentz - like force involving the relative superfluid velocity which is simply given by the gradient of the phase difference between the chiral superconducting components . f.s.n . and i.e. acknowledge the deutsche forschungsgemeinschaft ( dfg ) for the financial support via the collaborative research center sfb tr 12 . a.s .
acknowledges support from the research council of norway , grant nos .
205591/v20 and 216700/f20 , as well as support from cost action mp-1201 `` novel functionalities through optimized confinement of condensates and fields '' .
ie also acknowledges the financial support of the ministry of education and science of the russian federation in the framework of increase competitiveness program of nust misis ( n 22014015 ) . c. l. kane and e. j. mele , phys .
lett . * 95 * , 226801 ( 2005 ) ; b. a. bernevig , t. l. hughes , s .- c .
zhang , science * 314 * , 1757 ( 2006 ) ; l. fu , c. l. kane , phys .
b * 76 * , 045302 ( 2007 ) ; h. zhang , c .- x . liu , x .-
qi , x. dai , z. fang , s .- c .
zhang , nat .
phys . * 5 * , 438 ( 2009 ) .
m. knig , s. wiedmann , c. brne , a. roth , h. buhmann , l. w. molenkamp , x .-
qi , s .- c .
zhang , science * 318 * , 766 ( 2007 ) ; d. hsieh , d. qian , l. wray , y. xia , y. s. hor , r. j. cava , and m. z. hasan , nature * 452 * , 970 ( 2008 ) ; y. xia , d. qian , d. hsieh , l. wray , a. pal , h. lin , a. bansil , d. grauer , y. s. hor , r. j. cava , m. z. hasan , nat .
phys . * 5 * , 398 ( 2009 ) ; y. l. chen , j. g. analytis , j .- h .
chu , z. k. liu , s .- k .
mo , x. l. qi , h. j. zhang , d. h. lu , x. dai , z. fang , s. c. zhang , i. r. fisher , z. hussain , z .- x .
shen , science * 10 * , 178 ( 2009 ) . | we demonstrate that the higgs mechanism in three - dimensional topological superconductors exhibits unique features with experimentally observable consequences .
the higgs model we discuss has two superconducting components and an axion - like magnetoelectric term with the phase difference of the superconducting order parameters playing the role of the axion field . due to this additional term , quantum electromagnetic and phase fluctuations
lead to a robust topologically non - trivial state that holds also in the presence of interactions . in this sense , we show that the renormalization flow of the topologically nontrivial phase can not be continuously deformed into a topologically non - trivial one .
one consequence of our analysis of quantum critical fluctuations , is the possibility of having a first - order phase transition in the bulk and a second - order phase transition on the surface .
we also explore another consequence of the axionic higgs electrodynamics , namely , the anomalous hall effect . in the low frequency london regime an anomalous hall effect is induced in the presence of an applied electric field parallel to the surface .
this anomalous hall current is induced by a lorentz - like force arising from the axion term , and it involves the relative superfluid velocity of the superconducting components .
the anomalous hall current has a negative sign , a situation reminiscent of , but quite distinct in physical origin from the anomalous hall effect observed in high-@xmath0 superconductors .
in contrast to the latter , the anomalous hall effect in topological superconductors is non - dissipative and occurs in the absence of vortices . |
the realization of bose - einstein condensates ( bec)@xcite , achieved by taking dilute alkali gases to ultra low temperatures @xcite is certainly among the most exciting recent experimental achievements in physics .
since then , investigations dedicated to the comprehension of new phenomena associated to this state of matter as well as its properties have flourished , either in the experimental or theoretical domains .
the fast increasing of the control and production of bose - einstein condensates ( becs ) in different geometries has permitted the study of these systems in different physical situations .
the fragmentation of a bec to produce a josephson junction @xcite using becs opened the possibility to study the quantum tunnelling of the atoms across a barrier between the condensates @xcite .
using superposition of light in different direction is possible to create any arbitrary trapping configuration as for example a ring or a superconducting quantum interference devices ( squid ) with an atom bec @xcite .
another experimental realization of fragmentation of becs is the two - legs bosonic ladder to study chiral current and meissner effect @xcite .
there is also the possibility to produce optical lattices in one - dimension ( 1d ) , two - dimensions ( 2d ) and three - dimensions ( 3d ) using one , two or three orthogonal standing waves @xcite .
these experimental realisations have opened the possibility to introduce new models that permit the study of the tunnelling of the atoms between the becs .
some of these models are exactly solvable by the algebraic bethe ansatz method @xcite and this open the possibility to take in account quantum fluctuations that allows us go beyond the results obtained by mean field approximations
. this fruitful approach can furnish some new insights in this area , and contribute as well to the increasingly interesting field of integrable systems itself @xcite .
the algebraic formulation of the bethe ansatz , and the associated quantum inverse scattering method ( qism ) , was primarily developed in @xcite .
the qism has been used to unveil properties of a considerable number of solvable systems , such as , one - dimensional spin chains , quantum field theory of one - dimensional interacting bosons @xcite and fermions @xcite , two - dimensional lattice models @xcite , systems of strongly correlated electrons @xcite , conformal field theory @xcite , integrable systems in high energy physics @xcite and quantum algebras ( deformations of universal enveloping algebras of lie algebras ) @xcite . for
a pedagogical and historical review see @xcite .
more recently solvable models have also showed up in relation to string theories ( see for instance @xcite ) .
remarkably it is important to mention that exactly solvable models are recently finding their way into the lab , mainly in the context of ultracold atoms @xcite but also in nuclear magnetic resonance ( nmr ) experiments@xcite turning its study as well as the derivation of new models an even more fascinating field .
i am considering here the bosonic multi - states lax operator introduced by the author in @xcite that permits to solve a family of models of fragmented becs coupled by josephson tunnelling .
this lax operator is a generalization of the bosonic lax operator in @xcite , where a lax operator is defined for a single canonical boson operator , but instead of a single operator we choose a linear combination of independent canonical boson operators .
the paper is organized as follows . in section 2
, i will review briefly the algebraic bethe ansatz method and present the lax operators and the transfer matrix for both models . in section 3 , i present a generalized model with a family of models of fragmented becs coupled by josephson tunnelling and show their solutions . in section 4
, i summarize the results .
in this section we will shortly review the algebraic bethe ansatz method and present the transfer matrix used to get the solution of the models @xcite .
we begin with the @xmath0-invariant @xmath1-matrix , depending on the spectral parameter @xmath2 , @xmath3 with @xmath4 , @xmath5 and @xmath6 . above
, @xmath7 is an arbitrary parameter , to be chosen later .
it is easy to check that @xmath8 satisfies the yang - baxter equation @xmath9 where @xmath10 denotes the matrix acting non - trivially on the @xmath11-th and the @xmath12-th spaces and as the identity on the remaining space .
next we define the monodromy matrix @xmath13 , @xmath14 such that the yang - baxter algebra is satisfied @xmath15 in what follows we will choose a realization for the monodromy matrix @xmath16 to obtain solutions of a family of models for multilevel two - well bose - einstein condensates . in this construction , the lax operators @xmath17 have to satisfy the relation @xmath18 then , defining the transfer matrix , as usual , through @xmath19 it follows from ( [ rtt ] ) that the transfer matrix commutes for different values of the spectral parameter ; i. e. , @xmath20=0 , \;\;\;\;\;\;\ ; \forall \;u,\;v.\ ] ] consequently , the models derived from this transfer matrix will be integrable .
another consequence is that the coefficients @xmath21 in the transfer matrix @xmath22 , @xmath23 are conserved quantities or simply @xmath24-numbers , with @xmath25 = 0 , \;\;\;\;\;\;\
; \forall \;j,\;k.\ ] ] if the transfer matrix @xmath22 is a polynomial function in @xmath2 , with @xmath26 , it is easy to see that , @xmath27 for the standard bosonic operators satisfying the canonical commutation relations @xmath28 = [ \hat{p}_{i},\hat{q}_{j } ] = 0 , \qquad [ \hat{p}_{i},\hat{q}_{j}^{\dagger } ] = \delta_{pq}\delta_{ij}\hat{i},\ ] ] @xmath29= + \hat{p}_{j}^{\dagger}\delta_{pq}\delta_{ij } , \qquad [ \hat{n}_{pi},\hat{q}_{j}]= -\hat{p}_{j}\delta_{pq}\delta_{ij},\ ] ] with @xmath30 , @xmath31 and @xmath32 , we have the following lax operators , @xmath33 and @xmath34 if the conditions , @xmath35 and @xmath36 , are satisfied .
the above lax operators satisfy the equation ( [ rll ] ) .
in this section i present some applications of the lax operators ( [ l2 ] ) and ( [ l3 ] ) for models with different number of becs @xmath37 and @xmath38 .
the generalized hamiltonian is , @xmath39 the parameters , @xmath40 , describe the atom - atom @xmath41-wave scattering between the atoms in the respective becs , the @xmath42 parameters are the relative external potentials between the becs and @xmath43 are the energies in the becs .
the parameters @xmath44 are the tunnelling amplitudes .
the operators @xmath45 are the number of atoms operators .
the labels @xmath46 and @xmath47 stand for the becs @xmath37 and @xmath38 with @xmath48 and @xmath49 .
we just remark that @xmath50 .
the becs are coupled by josephson tunnelling and the total number of atoms , @xmath51 , is a conserved quantity , @xmath52 = 0 $ ] .
the state space is spanned by the base @xmath53 and we can write each vector state as @xmath54 where @xmath55 is the vacuum vector state in the fock space . we can use the states ( [ state1 ] ) to write the matrix representation of the hamiltonian ( [ h1 ] ) . the dimension of the space increase very fast when we increase @xmath56 , @xmath57 where @xmath58 is the total number of becs in the system and @xmath56 is a constant @xmath24-number , @xmath59 . in the case where we have only two becs @xcite ( one @xmath60 and one @xmath61 )
the dimension is @xmath62 . in the figs .
( [ gf1 ] ) and ( [ gf2 ] ) we show some graphs for different values of @xmath63 and @xmath64 .
the balls with their respective labels are representing the condensates and the tubes are representing the tunnelling of the atoms between the respective condensates . [ cols="^,^ " , ] now we use the co - multiplication property of the lax operators to write , @xmath65 following the monodromy matrix ( [ monod ] ) we can write the operators , @xmath66 taking the trace of the operator ( [ lh2 ] ) we get the transfer matrix @xmath67 from ( [ c14b ] ) we identify the conserved quantities of the transfer matrix ( [ tu2 ] ) , @xmath68 @xmath69 @xmath70 we can rewrite the hamiltonian ( [ h1 ] ) using these conserved quantities @xmath71 with the following identification for the parameters @xmath72 @xmath73 @xmath74 @xmath75 @xmath76 the hamiltonian ( [ h1 ] ) is related with the transfer matrix ( [ tu2 ] ) by the equation , @xmath77 we use as pseudo - vacuum the product state , @xmath78 with @xmath79 denoting the fock vacuum state for the becs @xmath80 and @xmath81 denoting the fock vacuum state for the becs @xmath82 , for @xmath83 and @xmath84 . for this pseudo - vacuum
we can apply the algebraic bethe ansatz method in order to find the bethe ansatz equations ( baes ) , @xmath85 the eigenvectors @xcite @xmath86 of the hamiltonian ( [ h1 ] ) or ( [ h4 ] ) and of the transfer matrix ( [ tu2 ] ) are @xmath87 and the eigenvalues of the hamiltonian ( [ h1 ] ) or ( [ h4 ] ) are , @xmath88 where the @xmath89 are solutions of the baes ( [ bae2 ] ) and @xmath56 is the total number of atoms .
we can choose arbitrarily the spectral parameter @xmath2 . choosing @xmath90 , for example , we can write the baes ( [ bae2 ] ) in the limit @xmath91 as just one equation @xmath92 with @xmath93 if the bethe roots
@xmath89 are real numbers , the bae ( [ bae3 ] ) is the equation of a @xmath56-dimensional sphere of radii @xmath94 and center in @xmath95 in this limit and with @xmath96 we can write the eigenvalues as @xmath97
i have solved a family of fragmented bose - hubbard models using the multi - states boson lax operators introduced by the author in @xcite .
these models can be considered as graphs , with the becs in the vertices and the edge representing the tunnelling between the respective becs .
the graphs can appear in one - dimension ( 1d ) , two - dimension ( 2d ) and in three - dimension ( 3d ) .
when we increase the number of becs we get a ring of becs @xmath37 and a chain of becs @xmath38 in the center of the ring .
we can consider the becs identical or different .
i have showed that in the limit @xmath91 , if the bethe roots are all real numbers , they are on a @xmath56-dimensional sphere .
the author acknowledge capes / faperj ( coordenao de aperfeioamento de pessoal de nvel superior / fundao de amparo pesquisa do estado do rio de janeiro ) for financial support .
10 kulish p p and sklyanin e k , _ integrable quantum field theories : proceedings of the symposium held at tvrminne , finland - lecture notes in physics _ editor : j. hietarinta and c. montonen , * 151 * , springer - verlag , berlin , ( 1982 ) 61 .
takhtajan l a , _ quantum groups : proceedings of the 8th international workshop on mathematical physics held at the arnold sommerfeld institute , clausthal , frg - lecture notes in physics _ ,
editor : h. -d .
doebner and j. -d .
hennig , * 370 * , springer - verlag , berlin , ( 1990 ) 3 .
essler f h l and korepin v e , _ exactly solvable models of strongly correlated electrons _ , world scientific , singapore , ( 1994 ) .
essler f h l , frahm h , ghmann f , klmper a and korepin v e , _ the one - dimensional hubbard model _ , cambridge university press , cambridge , ( 2005 ) .
bazhanov v , lukyanov s and zamolodchikov a b , _ commun .
_ * 177 * ( 1996 ) 381 .
lipatov l , _ jetp lett .
_ * 59 * ( 1994 ) 596 .
jimbo m , _ field theory , quantum gravity and strings : proceedings of a seminar series held at daphe , observatoire de meudon , and lpthe , universit pierre et marie curie , paris - lecture notes in physics _
, editor : h. j. de vega and n. snchez , * 246 * , springer - verlag , berlin , ( 1986 ) 335 . | i present the exact solution of a family of fragmented bose - hubbard models and represent the models as graphs in one - dimension , two - dimensions and three - dimensions with the condensates in the vertices .
the models are solved by the algebraic bethe ansatz method . |
quantum communications represent very important part of a rapidly developing research area called quantum information processing @xcite .
communications networks are nowadays an indispensable technology allowing people to transmit information quickly over large distances . in classical communications networks
classical physics laws are used for their construction .
the recent research in quantum physics and quantum information suggests that quantum laws of nature can provide significant improvement of capabilities of communications devices @xcite , while using similar resources and keeping similar network architecture @xcite .
not surprisingly , a significant amount of theoretical and experimental research has been dedicated to the concept of quantum communications networks @xcite .
most notable result of this effort is a number of protocols for quantum cryptography @xcite , which is a method for unconditionally secure transmission of information using various quantum properties of information carriers .
quantum communications networks can benefit from purely quantum effects such as entanglement or probabilistic nature of measurement .
the more complex both the classical and quantum networks are , the more pronounced is the need for correct routing of the signal from its source to its intended destination @xcite .
classical routers are well known ingredient of classical networks allowing to direct signal information according to control information ( e.g. ip address ) @xcite .
the analogy between classical and quantum networks suggests that complex quantum networks would also require elaborate routing protocols .
this need is even more pronounce since in contrast to classical information , one can not perfectly duplicate an unknown qubit of quantum information @xcite .
however , approximate cloning is possible and has been intensely studied both theoretically and experimentally over the last decades @xcite resulting in , e.g. , establishing and implementing optimal state - dependent cloning for a wide class of qubit distributions @xcite . nevertheless the impossibility of perfect cloning prevents using the concept of multi - directional broadcast ( known from classical networks ) in a quantum network . in this paper
we address the problem of designing a quantum router ( see conceptional scheme in fig . [
fig : concept ] ) .
we consider the platform of individual photons and linear optics because of its experimental accessibility and also because of the particular suitability of light for information transmission @xcite .
the router has to fulfil four requirements to be suitable for quantum communications networks : ( i ) both the signal and control information have to be stored in quantum objects ( qubits ) , therefore routers using classical information to route quantum signal are considered only semi - quantum routers .
( ii ) the signal information is unchanged under the routing operation , the degree of freedom used to store the signal qubit information has to be kept undisturbed .
( iii ) the router has to be able to route the signal into a coherent superposition of both output modes .
( iv ) the router has to work without any need for post - selection on the signal output .
if the router is probabilistic , successful operation can be identified by detection on the control state .
( v ) to optimize the resources of the quantum network , only individual control qubit is required to direct one signal qubit
. there has been several schemes for quantum router already proposed , some of them habe been experimentally implemented , but none of them however meets all the five requirements defined in the above paragraph .
the first group of these proposals uses light - matter interaction in order to achieve quantum routing @xcite .
such interaction is however often very challenging for experimental implementation .
the second group of the proposals considers solely the platform of optical interactions to accomplish the routing .
there are some proposals for semi - quantum router , where the control information is classical using intensive light pulses @xcite .
there are also proposals where the control information is quantum , but the signal state collapses in the router so the requirement ( ii ) is not met @xcite .
recently we have proposed a fully quantum router using only linear optics @xcite .
this device however does not meet the requirement ( v ) since it requires two quantum bits to control routing of one single qubit of signal information .
in this section we describe an all - linear - optical quantum router that meets all the five requirements mentioned in the introduction .
the router makes use of three quantum gates : controlled - phase gate ( c - phase gate ) @xcite , quantum non - demolition presence detection gate ( qnd gate ) @xcite and programmable - phase gate ( ppg ) @xcite . in order for the setup to work completely without need for signal post - selection , both the qnd and the detector d
have to be equipped with photon - number resolving detectors . without them ,
the router still works , but does not fulfil the requirement ( iv ) .
linear - optical schemes of all of these gates are already published and with the exception of the qnd gate also already tested experimentally .
the reader is encouraged to get more details in the cited papers .
[ fig : scheme ] shows the scheme for linear - optical implementation of the quantum router .
we propose using two degrees of freedom of individual photons : ( 1 ) polarization encoding is used to store both the signal and control qubits , ( 2 ) path ( spatial mode ) encoding is used for the routing operation , the signal photon is routed into superposition of two output ports . and @xmath0 half and quarter wave plate , ndf neutral density filter , c - phase controlled phase gate , qnd
quantum non - demolition detector , ppg programmable - phase gate , d / a polarization analysis ( for more details see ref .
@xcite ) . ]
signal qubit enters the setup using the input port @xmath1 while the control qubit enters the router using port @xmath2 .
let us assume the signal qubit takes the form of a general quantum polarization state @xmath3 and @xmath4 denote the states of horizontal and vertical linear polarizations and @xmath5 . for reasons apparent later , it is suitable to parametrize the state of the control qubit by angles @xmath6 and @xmath7 @xmath8 after the signal qubit enters the router , it is subjected to the polarizing beam splitter ( pbs@xmath9 ) transmitting horizontally polarized light and reflecting light of vertical polarization . at this point
the state of both signal and control qubits reads @xmath10 where indexes denote spatial modes of the qubits . for better readability ,
let us now consider the evolution of both arms of the interferometer ( labelled in fig .
[ fig : scheme ] as and ) separately starting with the lower arm corresponding to the first two terms in eq .
( [ inputstate ] ) : the horizontally polarized signal photon undergoes a generalized hadamard transformation on a half - wave plate ( hwp ) rotated by -30 @xmath11 subsequently both the signal and control qubit photons enter the c - phase gate implemented using polarization dependent beam splitter pdbs of intensity transmissivities @xmath12 and @xmath13 for vertical and horizontal polarization respectively .
post - selecting only on the cases where there is one photon in signal and one photon in control mode , the transformation performed by the c - phase gate renders the investigated first two terms of ( [ inputstate ] ) to @xmath14 in the next step , the signal qubit undergoes hadamard transform also using a hwp yielding the terms @xmath15 as mentioned above , the c - phase gate is successful only when the signal and control qubits leave the gate by separate output ports . in order to post
- select only on such cases , the control qubit has to be subjected to qnd presence detection via the qnd gate .
the qnd presence detection requires an entangled pair of photons as a resource , but these photons are of a fixed quantum state , are generated locally solely for the purposes of the qnd gate and do not take other part in the quantum network .
the control qubit state does not change under the qnd gate , but success probability factor of @xmath16 is added to take into account the success probability of the qnd gate . more information about the qnd gate can be found in ref . @xcite . in the last step ,
the control qubit is subjected to a half - wave plate rotated at 22.5 yielding terms @xmath17 and then the control qubit alone impinge on the ppg composed just of a pbs .
the ppg gate heralds successful operation only if there is a control qubit with horizontal polarization at its input thus projecting the signal in the lower arm onto @xmath18 and in this form it impinges on the pbs@xmath19 .
now we examine the evolution of the second two terms in eq .
( [ inputstate ] ) corresponding to the propagation of the signal qubit by the upper arm : in this case the control qubit enters the c - phase gate alone rendering the investigated terms to @xmath20 again , we post - select only on those cases , where the control qubit leaves the c - phase gate by the mode @xmath2 .
this post - selection is again assured by the qnd gate which witnesses the control qubit presence with success probability of @xmath16 .
considering also the action of hwp we get @xmath21 subsequently the signal qubit undergoes a hadamard transformation yielding @xmath22 in the next step , both the signal and control qubits meet on the ppg gate s polarizing beam splitter . since we post - select only on the cases where there is exactly one photon at the output of mode @xmath2 , we continue considering only the terms @xmath23 as usual , the control photon impinge on the detector . depending on
the outcome of the polarizing detection measurement performed , the signal qubit collapses into @xmath24 when diagonally polarized control qubit was detected or @xmath25 when we observe anti - diagonally polarized control photon . in the later case
, we do apply a feed - forward consisting of a hwp placed at 0 causing @xmath26 and thus correcting the signal to the form of ( [ eq : signal_mode2 ] ) .
subsequent hadamard gate renders the signal to the form of @xmath27 before the signal impinges on the second polarizing beam splitter pbs@xmath19 , we introduce filtering by neutral density filter ndf of transmissivity @xmath28 to balance the amplitude with respect to the lower arm contribution described in ( [ eq : arm1 ] ) . after having both the terms of eq .
( [ inputstate ] ) evaluated , let us remind that the total state of the signal qubit reads @xmath29 and after being subjected to the pbs@xmath19 it takes the form of @xmath30 additional half - wave plate at 0 in first output corrects @xmath31 and polarization swap @xmath32 is applied to the second output by inserting a half wave plate at 45 yielding the final form of the signal state at the output of the router @xmath33 where one can clearly observe the routing operation .
the polarization state remains in the original form of ( [ signalstate ] ) , but the spatial degree of freedom is modified depending on the parameter @xmath6 of the control qubit .
the ratio between the amplitudes @xmath34 and @xmath35 depend straightforwardly on this parameter @xmath36 where we have introduced the routing ratio parameter @xmath37 in a similar manner as splitting ratio parametrization is introduced for ordinary beam splitters . since @xmath6 lies in the interval @xmath38 $ ] , the router can direct all the signal to the first output ( @xmath35 ) , to the second output ( @xmath34 ) or to any of their superposition .
the success probability @xmath39 does not depend on the signal state parameters @xmath40 and @xmath41 , but only on the ratio @xmath37 and therefore on the control qubit parameter @xmath6 .
one can easily find the relation between success probability and the routing parameter @xmath37 or the control qubit parameter @xmath6 @xmath42 it reaches maximum of @xmath43 for the @xmath44 and , on the other hand , it is minimized for @xmath45 to a value of @xmath46 .
the plot in fig .
[ fig : psucc ] shows success probability of routing as a function of the routing parameter @xmath37 .
it also depicts the relation between the routing parameter @xmath37 and the control qubit parameter @xmath6 .
note that neutral density filter of intensity transmissivity of @xmath47 can be placed to output port @xmath48 in order to equalize the success probability so it is completely control state independent and fixed to the value of @xmath46 . .
additionally the plot shows the relation between the routing parameter @xmath37 and the control qubit parameter @xmath6 . ]
( color online ) ( a ) linear - optical scheme of a tunable c - phase gate .
the interaction mach - zehnder interferometer replaces the fixed polarization dependent beam splitter pdbs in the original scheme .
the phase shift imposed by the gate is tuned by setting specific values of phase shift ( ps ) and losses ( ndf ) in this interferometer ( for more details see ref .
( b ) modified scheme of the router using two tunable c - phase gates.,title="fig : " ] ( color online ) ( a ) linear - optical scheme of a tunable c - phase gate .
the interaction mach - zehnder interferometer replaces the fixed polarization dependent beam splitter pdbs in the original scheme .
the phase shift imposed by the gate is tuned by setting specific values of phase shift ( ps ) and losses ( ndf ) in this interferometer ( for more details see ref .
( b ) modified scheme of the router using two tunable c - phase gates.,title="fig : " ] in the previous section we have only considered c - phase gate with a fixed phase shift of @xmath49 ( also known as the controlled - sign gate ) . in some cases , however , higher success probability can be achieved using a tunable c - phase gate that can be set to exercise a phase shift @xmath50 of any value in the range @xmath51 $ ] .
the reason for considering this tunable c - phase gate lies in the fact that the success probability of the gate is a function of its phase shift . in 2010
, konrad kieling and his colleagues @xcite discovered the success probability @xmath52 relation to the phase shift @xmath50 is the optimal success probability achievable using only linear optics and vacuum ancillae . ]
@xmath53^{-2}.\end{aligned}\ ] ] in order to make the c - phase gate tunable , one needs to replace the fixed polarization dependent beam splitter by a interaction mach - zehnder interferometer with tunable phase and losses ( see fig .
[ fig : tunable_cphase]a ) .
in a recent paper @xcite , we showed that if one does not seek this success probability to be input state independent , the c - phase gate can be generalized to perform the transformation @xmath54 ) and introduced @xmath55 . note that
in the first and fifth case , there is no photon entering the interaction interferometer , while in the second , third and sixth case there is exactly one photon entering this interferometer . in the fourth case , both the photons
are subjected to the interferometer .
the success probability thus becomes input state dependent , but higher in average . when using the tunable c - phase gate for routing , several modifications of the setup have to be put in place ( see fig .
[ fig : tunable_cphase]b ) : firstly , the ppg gate has to be replaced by a second c - phase gate set to the same phase shift as the first one .
this c - phase gate would exercise the same transformation as described by ( [ eq : cphase_general ] ) , but with indexes @xmath56 swapped .
secondly , the transformation ( hwp ) performed on the control qubit between the two interaction gates is removed , so is the ndf in upper arm . and
finally , a generalized transformation in both arms and before the signal enters the c - phase gates has to be recalculated .
this transformation reflecting the success probability of the c - phase gate reads @xmath57 in and @xmath58 in .
this transformation assures the signal state independence of the routing procedure . after incorporating these modifications
, one can proceed in exactly the same manner as in section ii to reveal that the signal state just before impinging on the pbs@xmath19 takes the form of @xmath59 |h_1\rangle \nonumber \\ & & + \frac{\alpha \mathrm{e}^{\imath \vartheta } \sin\theta\sqrt{a^3_c}}{2\sqrt{2 + 2a_c}}(1-\mathrm{e}^{\imath \varphi})|v_1\rangle\end{aligned}\ ] ] in the first arm and the form of @xmath60|v_2\rangle \nonumber \\ & & + \frac{\beta \mathrm{e}^{\imath \vartheta } \sin\theta\sqrt{a^3_c}}{2\sqrt{2 + 2a_c}}(1-\mathrm{e}^{\imath \varphi})|h_2\rangle\end{aligned}\ ] ] in the second arm .
( [ eq : tunable_arm1 ] ) and ( [ eq : tunable_arm2 ] ) indicate that the value of the phase shift @xmath50 imposed by the gate limits the routing ratio @xmath37 @xmath61 for the sake of readability , we limit to @xmath62 in the remaining part of this section . thus even for @xmath63 the routing ratio is bound by the relation @xmath64 where we have introduced the routing ratio limit @xmath65 for a given phase shift @xmath50 . using this definition
, we can formulate the range of achievable routing ratios for a given phase shift @xmath50 to be @xmath66 $ ] obtained when tuning monotonously the control qubit in the range @xmath67 $ ] .
note that , as expected , @xmath68 for @xmath69 and @xmath70 for @xmath49 .
( color online ) maximum and minimum success probability of the router using two tunable c - phase gates depicted as functions of the routing ratio limit .
the plot indicates that for sufficiently small values of routing ratio limit , the tunable c - phase gate offers better performance then the scheme devised in previous section ( maximum and minimum success probability of the previous scheme is depicted using black dashed lines at @xmath43 and @xmath46 ) .
the dotted line shows the success probability if standard state - independent c - phase gate is used instead of the generalized version . ]
we can also easily find the success probability of the router by calculating the norm of the output state ( [ eq : tunable_arm1 ] ) and ( [ eq : tunable_arm2 ] ) .
this success probability function depends on both @xmath50 and @xmath6 . for a fixed value of @xmath50 we can find its minimum of @xmath71{a^4_c}}{2 + 2a_c}\ ] ] always for @xmath63 . on the other hand , the maximum success probability is obtained for different values of @xmath6 depending on the phase shift @xmath50 @xmath72.\ ] ] both the success probability and the routing ratio limit
are function of the gate phase shift @xmath50 . to illustrate their mutual relation , we have plotted the success probability as a function routing limit in fig .
[ fig : tune_prob ] . in this figure , we show the maximum and minimum success probability together with the range between them for a given routing ratio limit . observing this plot , we conclude that for sufficiently small values of routing ratio limit ( e.g. @xmath73 ) the tunable c - phase gate offers increased success probability in comparison with the scheme proposed in previous section . on the other hand , for larger routing ratio limits ( closer to @xmath74 ) , the former scheme does better since it uses only one c - phase gate together with a more efficient ppg gate .
( color online ) scheme for multi - qubit routing , c denotes classical optical coupler . ]
so far we have only considered routing a single signal qubit using one control qubit . in principle , however , the device can route a chain of signal qubits making use of the same control qubit . the fig . [
fig : concept2 ] depicts the configuration for such multi - qubit routing . in this model ,
two c - phase gates are used as in the previous section , but the control qubit is not detected immediately . after interacting with the first signal qubit , the control qubit presence is verified in a qnd gate and then it is transferred back to the input of the router so it meets with the second signal qubit . after interacting with the last signal qubit , the control qubit is subjected to polarization analysis thus projecting the state of all signal qubits .
let us assume the state of @xmath75-th signal qubit to be @xmath76 thus the state of the entire system can be expressed as @xmath77 where the control qubit @xmath78 has been introduced as in previous sections .
the procedure described in the previous paragraph then renders the state to @xmath79 where out1 and out2 denote output ports of the router .
subsequent detection of the control qubit in @xmath80 basis projects the state to the required form of @xmath81 since the routing operation is probabilistic , the success probability of routing @xmath75 qubits decreases exponentially with @xmath75 .
considering the success probability of one run of the router to be @xmath82 ( as calculated in previous section ) , the @xmath75-qubit router will perform with success probability @xmath83 it is worth noting that , apart from routing , this procedure can be used to generate the so - called noon states @xcite .
these states of the form of @xmath84 , where @xmath85 denotes the number of photons , are useful for instance in quantum metrology @xcite or quantum litography @xcite .
we have presented an all - linear - optical scheme for a fully quantum router .
the router meets all the requirements presented in the introduction .
this concept is more practical then some of the previously presented routers making use of experimentally difficult light - matter interaction or unrealistic strengths of non - linear optical phenomena @xcite .
in contrast to previously published linear - optical schemes , our router provides genuine quantum routing being controlled by a qubit while not disturbing the state of the signal qubit ( in contrast to @xcite ) .
also it only requires one control qubit to route a single signal qubit making it more resource - efficient ( in contrast to @xcite ) .
the proposed scheme operates with success probabilities ranging between @xmath43 and @xmath46 depending on the control qubit state .
we also discuss optimization of the success probability that can be reached by using tunable c - phase gates .
the presented analysis shows how efficiency can be increased upto @xmath16 if only small amplitude of the signal is needed to be sent to output port 2 .
note also that detector efficiency just scales the success probability , but does not change fidelity of the output .
we have also provided a recipe for multi - qubit router , where the same control qubit is used to route general number of signal qubits . in this case , the success probability of the procedure scales exponentially with the number of routed signal qubits .
the authors gratefully acknowledge the support by the operational program research and development for innovations
european regional development fund ( project no . cz.1.05/2.1.00/03.0058 .
kl acknowledges the support by the czech science foundation ( project no .
13 - 31000p ) , kb acknowledges project no
. cz.1.07/2.3.00/30.0041 by the czech ministry of education , a acknowledges project no .
cz.1.07/2.3.00/20.0017 by the czech ministry of education and js acknowledges support by the project avoz10100522 by the institute of physics of cas . | all - linear - optical scheme for fully featured quantum router is presented .
this device directs the signal photonic qubit according to the state of one control photonic qubit . in the introduction
we formulate the list of requirements imposed on a fully quantum router .
then we describe our proposal showing the exact principle of operation on a linear - optical scheme .
subsequently we provide generalization of the scheme in order to optimize the success probability by means of a tunable controlled - phase gate . at the end , we show how one can modify the device to route multiple signal qubits using the same control qubit . |
the basic structure of the spin - dependent hexagonal optical lattice discussed in this paper is illustrated in fig .
[ fig:1 ] .
three laser - beams intersecting under an angle of @xmath0 and with each beam linearly polarized in the plane of intersection lead to the formation of local potential minima in a hexagonal structure . for neighboring sites along the vertices of the hexagonal lattice as for example indicated in fig .
[ fig:1]a by a dashed line , the local polarization alternates between @xmath1 and @xmath2 . as atoms in a light field experience a polarization - dependent ac stark shift , the potential at these @xmath1 and @xmath2 sites is different for different atomic zeeman substates labeled by @xmath3 .
the potential can be written as : @xmath4 where the polarization of the light field @xmath5 ( @xmath6 for pure @xmath1 and @xmath7 for pure @xmath2 polarizations ) is mapped onto a pseudo - magnetic field @xmath8 , @xmath9 the land @xmath10-factor and @xmath11 the bohr magneton ( for details see appendix ) .
the potential consists of a spin - independent part @xmath12 of hexagonal symmetry and a state - dependent super - lattice emerging from the local pseudo - magnetic field @xmath13 . according to the local polarization
, we denote the emerging triangular substructures as @xmath1 and @xmath2 lattice ( see fig .
[ fig:1]a ) . the hexagonal lattice can therefore also be regarded as a triangular lattice with a bi - atomic basis where the atoms occupy @xmath1 and @xmath2 sites as indicated in fig .
[ fig:1]a by green and red bullets . for typical experimental parameters , the mobility of the atoms
is predominantly restricted to the hexagonal channel structure connecting the local minima of @xmath12 .
one possible tunneling channel is indicated in fig .
[ fig:1]a by a dashed line .
note that the large central peak maxima ( @xmath14 ) prevent a direct diagonal tunneling through these maxima .
figure [ fig:1]b illustrates the different lattice potentials along a tunneling channel for the @xmath15 hyperfine ground - state manifold of @xmath16rb .
the @xmath17 hyperfine states exhibit a similar behavior with twice the modulation amplitude for @xmath18 ( compare eq .
( [ potential ] ) ) .
+ to study atomic quantum phases in the lattice , we load an ensemble of ultracold @xmath16rb ground - state atoms into the lattice .
this consists of either pure or specific compositions of different hyperfine- and magnetic zeeman - states . due to the additional spin - dependent spatial variation of the potential ,
atoms with @xmath19 are preferably confined to either the @xmath1 or the @xmath2 triangular lattice structure .
in contrast , atoms in the @xmath20 state will distribute homogeneously over both the @xmath1 and @xmath2 lattices ( fig .
[ fig:1]b ) . for convenience
, we refer to these different configurations as _ triangular _
( @xmath19 ) and _ hexagonal _ states ( @xmath20 ) .
this spin - dependent spatial distribution determines the structure factor of the lattice , which in turn determines the quasi - momentum distribution of the atoms .
we experimentally observe this difference in the structure factors by mapping the quasi - momentum spectrum to the spatial density distribution of the ensemble via time - of - flight images .
typical results for _ triangular _ and _ hexagonal _ initial states are shown in fig .
[ fig:1]c together with the corresponding theoretical prediction from a one - particle band structure calculation .
the observed second- to first - order interference peak ratio ( see fig . [
fig:1]c bottom ) is significantly larger for the _ hexagonal _ states as compared to
_ triangular _ configurations .
it typically ranges between two and four for our experimental parameters , which is well reproduced by calculations of the bloch functions and their quasi - momentum distribution for the corresponding spin - dependent lattice geometry .
to obtain a detailed understanding of the rich physics arising from different lattice occupations we have developed a novel method for state- and site - selective microwave spectroscopy .
it allows for _ in - situ _ investigations of the spatial spin - ordering in the hexagonal and triangular lattice structures .
it is suitable for studying different regimes ranging from the superfluid to the mott - insulating phase .
previously , a different kind of polarization - dependent radio - frequency spectroscopy method was used to study cubic lattices@xcite .
our method is based on the spatially varying transition frequencies between populated @xmath3-states to initially non - occupied @xmath21-states .
the potential energy difference between the initial and the final state is given by @xmath22 , where @xmath23 is an additionally applied homogeneous magnetic guiding field and @xmath24 is the @xmath16rb - hyperfine splitting .
the spectroscopy signal is obtained by detecting and normalizing the number of atoms in the respective @xmath3- and @xmath21-states after applying a microwave pulse , then releasing the sample from the lattice and performing stern - gerlach separation . in the following we discuss central results of the microwave spectroscopy method , first for single component samples , then for spin - mixtures .
typical data for both scenarios is shown in fig .
[ fig : microwavespectroscopy ] . + the spatial variation of the differential potential shift @xmath25 vanishes for transitions with @xmath26 .
thus a single component sample initialized in the @xmath27 state shows a single transition line when the state @xmath28 is addressed ( fig .
[ fig : microwavespectroscopy]a ) . when transferring atoms from the _ hexagonal _ state @xmath29 to a _
triangular _ final state ( @xmath30 ; the data in fig .
[ fig : microwavespectroscopy]c shows @xmath29 to @xmath31 for @xmath32 ) , the spectrum splits up into two lines according to the energy difference between the neighboring lattice sites . here
, the left ( right ) transition peak refers to the transition from the first band of the _ hexagonal _ state @xmath29 to the first ( second ) band of the _ triangular _ state @xmath31 where the atoms are mostly confined to the @xmath33 sites .
we measure the two transition frequencies @xmath34khz and @xmath35khz , where @xmath36 is the resonance frequency without 2d - lattice .
we find excellent agreement with the theoretical expectations of @xmath37khz and @xmath38khz obtained using an _ ab initio _ single - particle 2d band - structure calculation .
note that for the hexagonal lattice the bloch functions do not separate like it is the case for cubic lattice .
+ moreover , we compare the particle transfer efficiencies of the two transitions with the calculated predictions .
experimentally , the transfer efficiencies are determined evaluating the integrals of the corresponding transition peaks . for the measurement shown in fig .
[ fig : microwavespectroscopy]c we measure the transfer efficiency of the @xmath39-transition to be @xmath40 times larger than for the @xmath41-transition .
this is in very good agreement with the calculated ratio of the franck - condon overlaps between the initial state and the two final states resulting in a @xmath42 times larger transfer efficiency for the @xmath39-transition .
it also confirms the expected homogeneous density distribution of @xmath43 atoms at @xmath1 and @xmath2 sites .
therefore the spectroscopy allows to measure the relative occupations of the two sublattices @xmath1 and @xmath2 .
it can equally well be applied to study other initial spin - states e.g. a state of @xmath44 atoms .
figures [ fig : microwavespectroscopy]b and e show the @xmath44 to @xmath28 transition amplitudes for different lattice depths and clearly reveal the exclusive occupation of the @xmath2 sites for sufficiently deep lattices . for shallow lattices ,
the microwave transition frequency coincides with the one for a vanishing lattice potential .
a crossover can be observed at intermediate lattice depths .
the observed shift of @xmath45 for deeper lattices is reproduced by calculations within a harmonic approximation , whereas the signals for shallow and intermediate lattices can not easily be explained in the single - site approximation .
when initializing the atoms in @xmath46 , we observe the equivalent signals corresponding to the occupation of the @xmath1 sites and thus demonstrate the ability to actively force the generation of a magnetically ordered nel state with predefined magnetization by using a mixture of different spin - states , e.g. , @xmath44 with @xmath46 .
+ the experimental setup allows for the preparation of arbitrary spin - mixtures in the spin - dependent optical lattice .
the interplay of the interaction between different spin - states and the lattice potential has a crucial impact on the static and dynamic properties of the system .
particularly interesting is the case of a mixture of @xmath29 and @xmath44 atoms . while the @xmath44 atoms tend to localize in the @xmath2 sites , the @xmath29 atoms fill the hexagonal lattice homogeneously as it was shown above for pure atomic samples ( figs .
[ fig : microwavespectroscopy]a - c , e ) . in case of a mixture of these spin - components , the additional repulsive _
inter-_species interaction causes the @xmath44 atoms to imprint a periodic density modulation on the @xmath29 atoms .
experimentally we observe an almost vanishing transition amplitude of the peak corresponding to the @xmath2 sites as shown in figure [ fig : microwavespectroscopy]d .
comparison of the integrated overall transition probability with the one obtained for pure systems indicate that ( @xmath47)% of the @xmath20 atoms , initially sitting at @xmath2 sites , have been transferred to @xmath1 sites .
this crystalline order imprinted onto the @xmath29 atoms by the @xmath44 atoms can appear in two different scenarios : first , when the @xmath29 atoms are localized on the @xmath1 sites , the system forms an alternating mott - insulator . in this case , the mixture resembles a localized alternating spin - ordering .
second , when the @xmath29 atoms are still superfluid the system shows an off - diagonal long - range order .
together with the spatial modulation due to the interaction with the other spin - component , this indicates a supersolid like behavior@xcite . by varying the depth of the lattice both scenarios have been realized and indeed , we experimentally observe that the @xmath29 atoms were driven across the superfluid - to - mott - insulator transition referring to both cases . and * b * ) absolute value of the superfluid order parameter @xmath48 of @xmath29 atoms .
two different ensembles are considered ; a pure ensemble of @xmath29 atoms ( blue line ) and a 50:50 mixture with additional @xmath44 atoms ( black lines ) .
the solid ( dashed ) black line is the value at @xmath1 ( @xmath2 ) sites . in a pure system the values are identical at @xmath1 and @xmath2 sites ( blue line ) .
the particle number of each spin - state is fixed to two per unit cell.,scaledwidth=35.0% ] + our observations are strongly supported by a theoretical analysis using the mean - field gutzwiller approximation@xcite ( see appendix ) . in fig .
[ fig : ss ] we demonstrate that an equal mixture of @xmath29 and @xmath44 atoms results in a _ forced supersolid_. this is shown with the aid of the absolute value of the superfluid order parameter @xmath49 ( top ) and the occupation number @xmath50 ( bottom ) of the @xmath29 state at @xmath1 and @xmath2 sites assuming a filling of two particles per unit cell . for comparison
, we also include the calculations for a pure @xmath29 system . here , @xmath51 and @xmath52 as well as @xmath53 and @xmath54
coincide for all @xmath55 revealing the expected homogeneous atom distribution within the lattice .
this changes when @xmath44 atoms are admixed to the @xmath29 atoms . at already moderate lattice depths
the @xmath44 particles are localized at the @xmath2 sites . as a consequence
the @xmath29 component is repelled from these sites and the corresponding occupation number and superfluid order parameter become spatially modulated , which is characteristic for a supersolid .
moreover , we observe that the effective blocking of @xmath2 sites by @xmath44 particles forces the @xmath29 atoms to undergo the transition to the mott insulator already at weaker lattice depths than in the pure case .
hence an admixture of a second spin - component allows us to tune the critical point of the sf
mi transition as well as the spatial density modulation and superfluid order parameter .
transport properties play an important role in understanding , for example , the conductivity of solid - state systems . for the quantum optical counterparts of such systems , the optical lattices ,
these properties are essentially governed by the interplay of the on - site interaction @xmath56 with the the tunnel matrix element @xmath57 and are usually well described by the bose hubbard model@xcite ( see appendix ) . in optical lattice experiments ,
the ratio of @xmath58 is directly adjustable through the power of the lattice beams allowing to drive the ensemble from the superfluid to the mott - insulating state .
we use the interference pattern contrast ( visibility ) as an indicator for this transition@xcite .
results of such measurements are shown in figure [ fig:3]a for different single - component samples .
we compare the sf
mi transition of a _ hexagonal _ state @xmath29 with that of the _ triangular _ states @xmath44 and @xmath59 and observe that the interference contrast curves are considerably shifted with respect to one another .
+ we start by analyzing the situation for the @xmath44 and the @xmath59 states .
as these states experience an additional modulation of the hexagonal potential caused by @xmath60 , resonant tunneling is only possible to the six _ next - nearest neighbor sites _ in the lattice . due to this additional modulation ,
the tunneling barrier between the occupied lattice sites increases with @xmath61 ( compare eq .
( [ potential ] ) ) .
the sf mi transition is thus expected to occur at lower overall beam powers for increasing values of @xmath61 .
however , the two transitions are expected to nearly coincide when both are rescaled to the same modulation depth .
this is confirmed by the data shown in the inset of figure [ fig:3]a , where the interference contrast is plotted as a function of the effective one - dimensional tunneling barrier height . for the @xmath20 states ,
the situation is different : since in this case the optical potential is not additionally modulated , resonant tunneling takes place to the three _
nearest neighbors_. this leads to a strongly increased tunneling probability and therefore higher lattice beam powers are required for the sf
mi transition to occur .
interestingly , all three transitions nearly coincide when normalized to the height of the barrier , regardless of the effect of different hopping distances and possible number of accessible sites .
this discrepancy can be explained when considering in addition to _
nearest neighbor _ tunneling processes also _ second - nearest neighbor _ processes .
this is confirmed in calculations using the gutzwiller approach .
results of such calculations are shown in fig .
[ fig:3]b for the experimental configurations with filling factors ranging between two and four per lattice unit cell .
the numerical results show a good qualitative agreement with the observed experimental data .
+ the fact that the sf mi transition occurs at different lattice beam intensities for different values of @xmath62 adds an exciting new dimension to our lattice geometry : it opens the possibility to create spin - mixtures where a fully localized component is immersed in a superfluid bath .
experimentally , we study this configuration by loading @xmath44 atoms which preferably occupy the @xmath2 lattice and add to these the more localized state @xmath59 ( @xmath63 ) , which is strongly confined to the @xmath1 ( @xmath2 ) lattice . as figure [ fig:3]c shows for the case of a 50:50 mixture , a considerable shift in the sf
mi transition of the @xmath44 atoms is observed when the admixed atoms occupy @xmath1 sites , i.e. , when they are located in between the preferably occupied @xmath64 sites of @xmath44 . on the other hand ,
the transition remains practically unaltered when the more localized component occupies the same lattice sites as the superfluid core .
the pictograms in figure [ fig:3]c illustrate these three different cases .
the addition of a spin - crystal thus allows one to tailor the position of the sf mi transition depending on the sub - lattice it occupies .
+ the gutzwiller computations reproduce the shift of the sf mi transition ( see figure [ fig:3]d ) .
the observed changes in the tunneling probabilities can be ascribed to a modification of second order hopping processes : second order hopping of @xmath44 between the @xmath2 sites is only possible by passing through sites of the @xmath1 lattice .
however , in a @xmath44
@xmath59 mixture , these sites are occupied by @xmath59 atoms .
the interaction energy between the atoms increases the effective barrier and suppresses tunneling .
thus , the sf mi transition is shifted towards lower lattice laser beam intensities . by changing the relative composition of the mixture , the strength of this tunneling suppression can be modified .
+ further insight into the underlying physical processes is gained by considering the phase diagrams calculated in gutzwiller approximation .
figure [ fig:4]a shows the sf mi phase diagram for the experimental configurations discussed in figs . [ fig:3]c and
d. a clear modification of the mott - insulator lobes of a _ triangular _ state can be observed when strongly localized atoms are present in the complementary triangular sublattice(@xmath65 ) .
such a strong modification with respect to the pure case does not occur when both components occupy the same sites ( @xmath66 ) .
this perfectly agrees with our experimental observations discussed above . in fig .
[ fig:4]b we plot the sf
mi transition for a @xmath67 mixture for different inter - component interaction strengths @xmath68 between @xmath44 and @xmath59 ranging from the experimental value @xmath68 to @xmath69 .
such a situation of e.g. @xmath69 could be achieved experimentally by tuning the interactions with the help of feshbach resonances@xcite .
interestingly , the standard tendency is inverted in the hard core limit , when different components occupy different sites .
this implies that the sf
mi transition for higher mott orders occurs in the interesting regime that interpolates between the bose hubbard regime and an array of coupled rotors , similar to an array of josephson junctions@xcite .
in conclusion , we have presented a comprehensive experimental and theoretical study of a novel type of optical lattice , which exhibits hexagonal spatial symmetry .
the intriguing polarization structure of the lattice sites can cause a break up of the hexagonal structure for particular spin - states leading to the creation of a triangular , magnetic sublattice .
the nature of this lattice potential , combined with the possibility to precisely control system parameters , makes this system especially suitable for studying new quantum phase mixtures of superfluid and mott - insulating states .
+ we could show that a combined effect of repulsive interaction and the specific hexagonal lattice structure leads to modifications of @xmath58 , which is clearly a different situation compared to e.g. bose - fermi mixture experiments@xcite where higher order effects effectively modify @xmath57 and @xmath56 . + as promising further directions , entropy effects and entropic cooling @xcite can be studied in the hexagonal system .
particularly interesting , however , is the prospect of studying polarized or unpolarized fermi gases , or fermi bose mixtures in the hexagonal lattice geometry . at half filling , both of these polarized systems will mimic the physics of graphene@xcite , i.e. , exhibit dirac s dispersion relation , artificial relativistic effects , etc . introducing artificial gauge fields in such situations@xcite will lead to the occurrence of the anomalous quantum hall effect and a whole variety of exotic quantum phase transitions ( see for instance ref .
furthermore , by only changing the polarization of the laser beams creating the lattice , a spin - independent triangular lattice can be created@xcite . a bosonic gas in the triangular lattice
might then be used to mimic frustrated antiferromagnetism by employing a time - dependent lattice modulation @xcite .
we thank dirk - sren lhmann and andr eckardt for stimulating discussions .
moreover , we thank the deutsche forschungsgemeinschaft dfg for financial support within the forschergruppe for801 and the grk 1355 and support by the joachim herz stiftung .
support by the spanish mincin ( fis2008 - 00784 and consolider qoit ) , the alexander von humboldt foundation , caixa manresa , erc grant quagatua , and by the eu strep namequam are gratefully acknowledged .
* spin - dependent hexagonal optical lattice . * the light for the optical lattice is derived from a ti : sapphire laser running at @xmath70 nm . to realize the two - dimensional hexagonal lattice we interfere three laser - beams under an angle of @xmath71 in one plane .
the polarizations of the beams are linear and in the plane of the intersecting area and perpendicular to a homogeneous magnetic guiding field of @xmath72 which defines the quantization - axis . in order to compensate for phase - noise ,
the phase of each laser - beam is actively stabilized . to ensure two - dimensionality of the hexagonal structure , a perpendicular one - dimensional lattice with different frequency is applied .
typical potential depths of this 1d - lattice are @xmath73 .
the resulting two - dimensional , hexagonal lattice potential is given by @xmath74 + \cos\left ( { \mathbf{b}}_1 \cdot { \mathbf{x}}\right ) \nonumber \\ & + \cos\left({\mathbf{b}}_2 \cdot { \mathbf{x}}\right ) \ !
\big \ } - 2 \sqrt{3 } v_\mathrm{lat } \alpha \ , m_f \big
\ { \sin\left [ ( { \mathbf{b}}_1-{\mathbf{b}}_2)\cdot { \mathbf{x}}\right ] \nonumber \\ & + \sin\left ( { \mathbf{b}}_1 \cdot { \mathbf{x}}\right ) - \sin\left({\mathbf{b}}_2 \cdot { \mathbf{x}}\right ) \ ! \big \ } \nonumber \\ & \equiv v_{\mathrm{hex}}\left({\mathbf{x}}\right ) + m_f g_f \mu_{\mathrm{b } } b_{\mathrm{eff}}\left({\mathbf{x}}\right ) , \end{aligned}\ ] ] where @xmath75 and @xmath76 with @xmath77 are the reciprocal vectors of the hexagonal lattice .
the constant @xmath78 depends on the detuning of the lattice laser from the @xmath16rb resonances at 780/795 nm and is given for the experiments reported here ( @xmath79 ) @xmath80 .
the total potential is then given by @xmath81 .
the one - dimensional representation ( figure [ fig:1]b ) can be reduced to a sinusoidal potential with 319 nm period , modulated by a cosine function with 638 nm period .
the energy structure of the system can be readily calculated using an _ ab initio _ single - particle 2d band - structure calculation .
+ * preparation scheme .
* we start with a bose - einstein condensate of typically several @xmath82 atoms in the stretched state @xmath44 , which are confined in a nearly isotropic crossed dipole trap @xmath83 ) .
the preparation of the different pure and mixed spin - states is performed with aid of radio - frequency and/or microwave sweeps .
after the state preparation we ramp up the optical lattice within 80ms using an exponential ramp . within the ramping time the coherence between different spin - states is lost . +
* spectroscopy . *
we apply a microwave square - pulse of typically 2ms duration allowing for a sufficient resolution in frequency space . in order to minimize finite - state interactions ,
the intensity of the microwave pulse is adjusted such that typically only a small fraction of up to 20% of the atoms populate the final state .
to separate different spin - components during 27ms time - of - flight prior to absorption imaging , a stern - gerlach gradient field is applied . + + * bose hubbard model and gutzwiller mean - field ansatz . * for the computation of the theoretical phase diagrams ( figs . [ fig:3]b , d , and fig .
[ fig:4 ] ) the system is modeled by a bose hubbard model , which is known to give an essentially exact description of ultra - cold atoms in optical lattices@xcite .
the corresponding hamiltonian reads @xmath84 \nonumber \\ & + u^{\sigma}_{a , b } \,\hat n_{a , i } \hat n_{b ,
i } \big ) \big\},\end{aligned}\ ] ] where @xmath85 and @xmath86 denote the two different spin - components and @xmath1 and @xmath2 distinguish between the two sublattices .
the operator @xmath87 ( @xmath88 ) destroys an @xmath85 ( @xmath86 ) boson at site @xmath89 , and @xmath90 ( @xmath91 ) is the corresponding occupation operator .
angle brackets denote pairs of _ nearest neighbors_. the tunnel matrix element is denoted by @xmath92 , the on - site interaction between the spin - components @xmath93 and @xmath94 at @xmath95 site by @xmath96 .
local potentials are denoted by @xmath97 and the chemical potentials by @xmath98 .
the parameters @xmath97 , @xmath99 , and @xmath92 are extracted from the experimental setting . for the derivation of the on - site interaction @xmath56 we integrate over the gaussians which approximate well the wannier functions at the minima of the optical lattice .
we estimate the _
nearest neighbor _ tunneling @xmath57 from exact one - particle band - structure calculations of the full optical lattice . comparing this to a tight - binding hamiltonian
gives the desired tunneling coefficients .
+ to solve the model hamiltonian eq .
we have employed the standard gutzwiller mean - field approximation which is known to give good qualitative and quantitative account of the phase diagram in two and more dimensions and which reproduces the exact solution for @xmath100 and , in the thermodynamic limit , also for @xmath101.@xcite the superfluid order parameter is defined as the expectation value of the annihilation operator . in order to calculate the phase diagrams in fig .
[ fig:4 ] , we have chosen the chemical potential to be equal for both components , i.e. , @xmath102 . in the @xmath67 mixture this leads to an approximately 50:50 mixture , while in the @xmath103 mixture the @xmath44 atoms repel the @xmath63 strongly such that the occupation number can be heavily asymmetric . | periodicity is one of the most fundamental structural characteristics of systems occurring in nature .
the properties of these systems depend strongly on the symmetry of the underlying periodic structure . in solid state materials
for example the static and transport properties as well as the magnetic and electronic characteristics are crucially influenced by the crystal symmetry . in this context
, hexagonal structures play an extremely important role and lead to novel physics like that of carbon nanotubes or graphene . here
we report on the first realization of ultracold atoms in a spin - dependent optical lattice with hexagonal symmetry .
we show how combined effects of the lattice and interactions between atoms lead to a forced antiferromagnetic nel order when two spin - components localize at different lattice sites .
we also demonstrate that the coexistence of two components one mott - insulating and the other one superfluid leads to the formation of a forced supersolid .
our observations are consistent with theoretical predictions using gutzwiller mean - field theory . in recent years
, ultracold atoms in periodic potentials@xcite have been widely recognized as an important tool to simulate solid state systems and study their transport@xcite and magnetic properties@xcite .
different types of spin- and/or state - dependent lattices have been implemented and studied@xcite .
so far , most experiments with ultracold quantum gases have been carried out in lattices of cubic symmetry .
however , recent theoretical developments@xcite are aiming at systems with a hexagonal geometry .
in particular , carbon nanotubes@xcite , graphene@xcite and a large number of other carbon - based compounds show fascinating effects and exhibit particularly rich quantum phases@xcite . in this paper
we discuss the first realization of ultracold quantum gases in a hexagonal , spin - dependent optical potential and demonstrate how it can be used to tailor quantum phases of spin - mixtures and their dynamics .
we show that the combination of interactions between different spin - states and the spin - dependent lattice potential leads to novel quantum phases : a forced antiferromagnetic nel order when two spin - components localize in different sublattices , and a _ forced _ supersolid phase , with one spin - component being localized in one sublattice and inducing a density modulation on the other superfluid spin - component .
these phenomena are studied by exploiting a novel technique of state- and site - selective microwave - spectroscopy .
examining the impact on the onset of the superfluid to mott - insulator transition of the respective spin - state , we demonstrate furthermore that the mobility of particles in the lattice can be adjusted by immersing a well localized spin - crystal into a superfluid bath .
we calculate the corresponding phase diagrams within gutzwiller mean - field theory and find good agreement with the experimental results . |
numerical methods for partial differential equations ( pdes ) on stationary and evolving surfaces and for coupled bulk
surface pdes are under intensive research in the recent years .
surface finite element methods are all based on the fundamental paper of @xcite , further developed for evolving surface parabolic problems by @xcite .
high - order versions of various finite element methods for problems on a _ stationary surface _ have received attention in a number of publications previously .
we give a brief overview on this literature here : * the surface finite element method of @xcite was extended to higher order finite elements on stationary surfaces by @xcite .
some further important results for higher order surface finite elements were shown by @xcite . *
discontinuous galerkin method for elliptic surface problems were analysed by @xcite , and then extended to high - order discontinuous galerkin methods in @xcite . *
recently , the _ unfitted _ ( also called trace or cut ) finite element methods are investigated intensively , see for instance @xcite , @xcite and @xcite . a higher order version of the trace finite element method was analysed by @xcite . however , to our knowledge , there are no papers showing convergence of the _ high - order _ evolving surface finite element method for parabolic partial differential equations on _ evolving surfaces_. in this paper we extend the @xmath0 and @xmath1 norm convergence results of @xcite and @xcite to _ high - order _ evolving surface finite elements applied to parabolic problems on evolving surfaces with prescribed velocity . furthermore , convergence results for full discretisations using runge kutta methods , based on @xcite , and using backward difference formulae ( bdf ) , based on @xcite , are also shown . to prove high - order convergence of the spatial discretisation
three main groups of errors have to be analysed : * _ geometric errors _ , resulting from the appropriate approximation of the smooth surface .
many of these results carry over from @xcite by careful investigation of time dependencies , while others are extended from @xcite and @xcite , @xcite . *
_ perturbation errors of the bilinear forms _ , whose higher order version can be shown by carefully using the core ideas of @xcite .
* high - order estimates for the _ errors of a modified ritz projection _ , which was defined in @xcite .
these projection error bounds rely on the nontrivial combination of the mentioned geometric error bounds and on the well known aubin nitsche trick .
we further show convergence results for full discretisations using runge kutta and bdf methods .
the error estimates for runge kutta methods shown in @xcite and for bdf methods in @xcite are applicable without any modifications , since , the semidiscrete problem can be written in a matrix - vector formulation , cf .
@xcite , where the matrices have exactly the same properties as in the linear finite element case .
therefore , the fully discrete convergence results transfer to high - order evolving surface finite elements using the mentioned error estimates of the ritz map .
the implementation of the high - order method is also a nontrivial task .
the matrix assembly of the time dependent mass and stiffness matrices is based on the usual reference element technique .
similarly to isoparametric finite elements , the approximating surface is parametrised over the reference element , hence , all the computations are done there .
it was pointed out by @xcite , that the approach of @xcite requires _
explicit knowledge of the exact signed distance function to the surface @xmath2_. however , the signed distance function is only used in the analysis , but it is not required for the computations away from the initial time level . it is only used for generating the initial surface approximation . here we only consider linear parabolic pdes on evolving surfaces , however we believe our techniques and results carry over to other cases , such as to the cahn hilliard equation @xcite , to wave equations @xcite , to ale methods @xcite , @xcite , nonlinear problems @xcite and to evolving versions of bulk surface problems @xcite . furthermore , while in this paper we only consider evolving surfaces with prescribed velocity , many of the high - order geometric estimates of this paper are essential for the numerical analysis of parabolic problems where the _ surface velocity depends on the solution _ , cf .
@xcite .
the paper is organised in the following way : in section [ section : the problem ] we recall the basics of linear parabolic problems on evolving surfaces together with some notations based on @xcite .
section [ section : esfem ] deals with the description of higher order evolving surface finite elements based on @xcite .
section [ section : main result ] contains the time discretisations and states the main results of this paper : semidiscrete and fully discrete convergence estimates . in section [ section : geometric estimates ]
we turn to the estimates of geometric errors and geometric perturbation estimates . in section [ section : ritz ]
the errors in the generalised ritz map and in its material derivatives are estimated .
section [ section : error bounds ] contains the proof of the main results .
section [ section : implementation ] shortly describes the implementation of the high - order evolving surface finite elements . in section [ section : numerics ]
we present some numerical experiments illustrating our theoretical results .
let us consider a smooth evolving compact surface @xmath3 ( @xmath4 ) , @xmath5 , which moves with a given smooth velocity @xmath6 .
let @xmath7 denote the material derivative of the function @xmath8 , where @xmath9 is the tangential gradient given by @xmath10 , with unit normal @xmath11 .
we are sharing the setting of @xcite , @xcite .
we consider the following linear problem on the above surface , for @xmath12 , @xmath13 where for simplicity we set @xmath14 , but all our results hold with a non - vanishing inhomogeneity as well .
let us briefly recall some important concepts used later on , whereas in general for basic formulas we refer to @xcite .
an important tool is the green s formula on closed surfaces @xmath15 we use sobolev spaces on surfaces : for a smooth surface @xmath16 , for fixed @xmath17 $ ] , we define @xmath18 and analogously @xmath19 for @xmath20 , cf
. section 2.1 of @xcite .
finally , the space - time manifold is denoted by @xmath21 } { \gamma}\t\times\{t\}$ ] .
the variational formulation of this problem reads as : find @xmath22 such that @xmath23 holds for almost ever @xmath24 for every @xmath25 , and @xmath26 holds . for suitable @xmath27 existence and uniqueness results for
were obtained by theorem 4.4 in @xcite .
we define the high - order evolving surface finite element method ( esfem ) applied to our problem following @xcite and @xcite , @xcite .
we use simplicial elements and continuous piecewise polynomial basis functions of degree @xmath28 .
the smooth _ initial _ surface @xmath29 is approximated by a @xmath28-order interpolating discrete surface denoted by @xmath30 , with vertices @xmath31 , @xmath32 , and it is given as @xmath33 more details and the properties of such a discrete initial surface can be found in section 2 of @xcite .
then the nodes of the initial triangulation , sitting on the exact surface , are moved along the space - time manifold with the known velocity @xmath6 .
hence , the nodes @xmath34 define a triangulation also for @xmath35 , by @xmath36 the discrete surface @xmath37 remains to be an interpolation of @xmath16 for all times .
we always assume that the evolving elements @xmath38 are forming an admissible triangulation @xmath39 , with @xmath40 denoting the maximum diameter , cf .
section 5.1 of @xcite .
the discrete tangential gradient on the discrete surface @xmath41 is given by @xmath42 understood in an elementwise sense , with @xmath43 denoting the normal to @xmath37 , see @xcite , @xcite .
for every @xmath17 $ ] and for the @xmath28-order interpolation @xmath41 we define the finite element subspace of order @xmath28 , denoted by @xmath44 , spanned by the continuous evolving basis functions @xmath45 of piecewise degree @xmath28 , satisfying @xmath46 for all @xmath47 , therefore @xmath48 we interpolate the surface velocity on the discrete surface using the basis functions and denote it with @xmath49 . then the discrete material derivative is given by @xmath50 the key _
transport property _ derived in proposition 5.4 in @xcite , is the following @xmath51 the spatially discrete problem ( for a fixed degree @xmath28 ) then reads as : find @xmath52 such that @xmath53 with the initial condition @xmath54 being a suitable approximation to @xmath55 . later on
we will always work with a high - order approximation surface @xmath41 and with the corresponding high - order evolving surface finite element space @xmath44 ( with @xmath56 ) , therefore from now on we drop the upper index @xmath28 , unless we would like to emphasize it , or it is not clear from the context .
similarly to @xcite , the ode form of the above problem can be derived by setting @xmath57 into the semidiscrete problem , and by testing with @xmath58 ( @xmath59 ) , and using the transport property .
the spatially semidiscrete problem is equivalent to the following ode system for the vector @xmath60 , collecting the nodal values of @xmath61 , @xmath62 where the evolving mass matrix @xmath63 and the stiffness matrix @xmath64 are defined as @xmath65 for @xmath66 . for the error analysis
we need to compare functions on different surfaces .
this is conveniently done by the _ lift operator _ , which was introduced in @xcite and further investigated in @xcite , @xcite .
the lift operator maps a function on the discrete surface onto a function on the exact surface .
let @xmath67 be a @xmath28-order approximation to the exact surface @xmath68 . using the oriented distance function @xmath69 ( cf .
section 2.1 of @xcite ) , the lift of a continuous function @xmath70 is defined as @xmath71 where for every @xmath72 the point @xmath73 is uniquely defined via @xmath74 by @xmath75 we denote the function on @xmath67 whose lift is @xmath76 . in particular we will often use the space of lifted basis functions @xmath77
we are now formulate the convergence theorem for the semidiscretisation using high - order evolving surface finite elements .
this result is the higher order extension of theorem 4.4 in @xcite .
[ theorem : semidiscrete convergence ] consider the evolving surface finite element method of order @xmath28 as space discretisation of the parabolic problem .
let @xmath8 be a sufficiently smooth solution of the problem , and assume that the initial value satisfies @xmath78 then there exists @xmath79 such that for @xmath80 , the following error estimate holds for @xmath81 : @xmath82 the constant @xmath83 is independent of @xmath84 and @xmath85 , but depends on @xmath86 .
the proof of this result is postponed to a later section , after we have shown some preparatory results .
we apply a @xmath87-step backward difference formula ( bdf ) for @xmath88 as a discretisation to the ode system , coming from the esfem space discretisation of the parabolic evolving surface pde .
we briefly recall the @xmath87-step bdf method applied to the system with step size @xmath89 : @xmath90 where the coefficients of the method are given by @xmath91 , while the starting values are @xmath92 .
the method is known to be @xmath93-stable for @xmath94 and have order @xmath87 ( for more details , see chapter v. of @xcite ) . in the following result
we compare the fully discrete solution @xmath95 obtained by solving and the ritz map @xmath96 ( @xmath17 $ ] ) of the sufficiently smooth solution @xmath8 .
the precise definition of the ritz map is given later .
the following error bound was shown in @xcite for bdf methods up to order five .
see also @xcite .
[ theorem : bdf error estimates ] consider the parabolic problem , having a sufficiently smooth solution for @xmath97 .
couple the @xmath28-order evolving surface finite element method as space discretisation with time discretisation by a @xmath87-step backward difference formula with @xmath88 .
assume that the ritz map of the solution has continuous discrete material derivatives up to order @xmath98 .
then there exists @xmath99 , independent of @xmath40 , such that for @xmath100 , for the error @xmath101 the following estimate holds for @xmath102 : @xmath103 where the constant @xmath83 is independent of @xmath104 and @xmath105 , but depends on @xmath86 .
furthermore @xmath106 the @xmath107-norm , of the esfem residual @xmath108 , is defined as @xmath109 the most important technical tools in the proof of theorem [ theorem : bdf error estimates ] , and also in the proof of the bdf stability result lemma 4.1 in @xcite , are the ode formulation , and the key estimates first shown in lemma 4.1 in @xcite . in this result the following estimates are shown : there exist @xmath110 such that , for @xmath111 , @xmath112 the proof of these bounds only involve basic properties of the mass and stiffness matrix @xmath63 and @xmath64 , hence it is independent of the order of the basis functions .
therefore , the high - order esfem versions of the above inequalities also hold .
therefore the original results of @xcite also hold here .
we are now in the position to formulate one of the main results of this paper , which yields optimal - order error bounds for high - order finite element semidiscretisation coupled to bdf methods up to order five applied to an evolving surface pde .
[ theorem : esfem bdf error ] consider the evolving surface finite element method of order @xmath28 as space discretisation of the parabolic problem , coupled to the time discretisation by a @xmath87-step backward difference formula with @xmath88 .
let @xmath8 be a sufficiently smooth solution of the problem , and assume that the starting values are satisfying ( with @xmath113 ) @xmath114 then there exists @xmath79 and @xmath99 , such that for @xmath80 and @xmath100 , the following error estimate holds for @xmath102 : @xmath115 the constant @xmath83 is independent of @xmath84 and @xmath85 , but depends on @xmath86 .
the proof of this result is also postponed to a later section , after we have shown some preparatory results .
[ remark : runge kutta extension ] we remark here that an analogous fully discrete convergence result is readily available for algebraically stable implicit runge kutta methods ( such as the radau iia methods ) , since the runge kutta analog of theorem [ theorem : bdf error estimates ] have been proved in @xcite .
the combination of this result with our high - order semidiscrete error bounds ( theorem [ theorem : res bound ] ) proves the runge kutta analog of theorem [ theorem : esfem bdf error ] .
in this section we present further notations and some technical lemmas that will be used later on in the proofs leading to the convergence result .
these estimates are high - order analogs of some previous results proved in @xcite , @xcite , @xcite , @xcite and @xcite . in the following we state and prove estimates for the errors resulted by the geometric surface approximation .
[ lemma : equivalence of discrete norms ] let @xmath116 with lift @xmath117 .
then the discrete and continuous @xmath118- and sobolev norms are equivalent , independently of the mesh size @xmath40 .
for instance , there is a constant @xmath119 such that for all @xmath40 : @xmath120 we now turn to the study of some geometric concepts and their errors . by @xmath121
we denote the quotient between the continuous and discrete surface measures , @xmath122 and @xmath123 , defined as @xmath124 .
further , we recall that @xmath125 and @xmath126 are the projections onto the tangent spaces of @xmath68 and @xmath67 , respectively .
we further set , from @xcite , @xmath127 where @xmath128 ( @xmath129 ) is the ( extended ) weingarten map . using this notation and , in the proof of lemma 5.5 @xcite
it is shown that @xmath130 for these quantities we show some results analogous to their linear esfem version showed in lemma 5.4 in @xcite , proposition 2.3 in @xcite , or lemma 6.1 in @xcite . later on the following estimates
will play a key technical role .
[ lemma : geometric est ] for @xmath41 and @xmath131 as above , we have the geometric approximation estimates : @xmath132 with constants depending only on @xmath133 , but not on @xmath40 or @xmath134 .
the first three bounds were shown in proposition 2.3 of @xcite for the stationary case .
noting that the constants only depend on @xmath133 these inequalities are shown .
the last four bounds are simply the higher order extensions of the corresponding estimates in lemma 6.1 of @xcite .
they can be proved in the exact same way using the bounds of the first three estimates .
the following result gives estimates for the error in the interpolation .
our setting follows that of section 2.5 of @xcite .
let us assume that the surface @xmath68 is approximated by the interpolation surface @xmath37 .
then for any @xmath135 , there is a unique @xmath28-order surface finite element interpolation @xmath136 , furthermore we set @xmath137 .
[ lemma : interpolation error ] let @xmath138 such that @xmath135 for @xmath5 .
there exists a constant @xmath119 depending on @xmath133 , but independent of @xmath40 and @xmath134 such that , for @xmath5 , @xmath139 we distinguish the special case of a linear surface finite element interpolation on @xmath41 . for @xmath140 the linear surface finite element interpolant
is denoted by @xmath141 , and it satisfies , with @xmath119 , @xmath142 note that for @xmath143 the underlying approximating surface @xmath41 is still of high order . for @xmath144 , @xmath145 and @xmath143 simply coincide .
the upper @xmath28 indices are again dropped later on .
following @xcite and @xcite we define the velocity of the lifted material points , denoted by @xmath146 , and the corresponding discrete material derivative @xmath147 . for arbitrary @xmath148 , with @xmath149 , cf .
, we have @xmath150 hence for @xmath151 , see @xcite , @xmath152 following section 7.3 of @xcite , we note that @xmath153 is the normal component of @xmath154 , and that the other terms are tangent to @xmath16 , hence @xmath155 it is also important to note that @xmath156 , cf .
@xcite .
the discrete material derivative of the lifted points on @xmath16 read @xmath157 the lifted basis functions also satisfy the transport property @xmath158 to prove error estimates for higher order material derivatives of the ritz map , we need high - order bounds for the error between the continuous velocity @xmath6 and the discrete velocity @xmath146 .
we generalise here lemma 5.6 in @xcite and lemma 7.3 in @xcite to the high - order case .
[ lemma : velocity estimate ] for @xmath159 , there exits a constant @xmath160 depending on @xmath133 , but independent of @xmath134 and @xmath40 , such that @xmath161 we follow the steps of the original proofs from @xcite and @xcite .
\(a ) for @xmath162 . using the definition and the fact @xmath163 we have @xmath164 where we used the interpolation estimates , lemma [ lemma : geometric est ] , and the boundedness of the other terms .
for the gradient estimate we use the fact that @xmath165 and the geometric bounds of lemma [ lemma : geometric est ] : @xmath166 \(b ) for @xmath167 , we use the transport property and again lemma [ lemma : geometric est ] : @xmath168 again using @xmath165 and the geometric bounds of lemma [ lemma : geometric est ] : @xmath169 \(c ) for @xmath170 the proof uses similar arguments .
we use the time dependent bilinear forms defined in @xcite : for arbitrary @xmath171 and for their discrete analogs for @xmath172 @xmath173 m(t;z,{\varphi } ) & = \int_{{\gamma}(t)}\!\!\!\ ! z { \varphi } , \\
a(t;z,{\varphi } ) & = \int_{{\gamma}(t)}\!\!\!\ ! { \nabla_{\gamma}}z \cdot { \nabla_{\gamma}}{\varphi } , \\ g(t;v;z,{\varphi } ) & = \int_{{\gamma}(t)}\!\!\!\ ! ( { \nabla_{\gamma}}\cdot v ) z{\varphi } , \\ b(t;v;z,{\varphi } ) & = \int_{{\gamma}(t)}\!\!\!\ ! { \mathcal{b}}(v ) { \nabla_{\gamma}}z \cdot { \nabla_{\gamma}}{\varphi } , \end{aligned } \qquad \begin{aligned}[c ] m_h(t;z_h,\phi_h ) & = \int_{{\gamma}_h(t)}\!\!\!\ ! z_h \phi_h , \\
a_h(t;z_h,\phi_h ) & = \int_{{\gamma}_h(t)}\!\!\!\ ! { \nabla_{\gamma_h}}z_h \cdot { \nabla_{\gamma_h}}\phi_h , \\ g_h(t;v_h;z_h,\phi_h ) & = \int_{{\gamma}_h(t)}\!\!\!\ ! ( { \nabla_{\gamma_h}}\cdot v_h ) z_h \phi_h , \\
b_h(t;v_h;z_h,\phi_h ) & = \int_{{\gamma}_h(t)}\!\!\!\ ! { \mathcal{b}}_h(v_h ) { \nabla_{\gamma_h}}z_h \cdot { \nabla_{\gamma_h}}\phi_h , \end{aligned}\ ] ] where the discrete tangential gradients are understood in a piecewise sense , and with the tensors given as @xmath174 for @xmath175 . for more details
see lemma 2.1 in @xcite ( and the references in the proof ) , or lemma 5.2 in @xcite .
we will omit the time dependency of the bilinear forms if it is clear from the context .
the following transport equations , especially the one with discrete material derivatives on the continuous surface , are of great importance during the defect estimates later on .
[ lemma : transport prop ] for any @xmath176 we have @xmath177 the same formulas hold when @xmath147 and @xmath146 are replaced with @xmath178 and @xmath6 , respectively .
similarly , in the discrete case , for arbitrary @xmath179 we have @xmath180 the following estimates are the most important technical results of this paper .
later on , they will play a crucial role in the defect estimates .
we note here that these results extend the first order esfem theory of @xcite ( for the first three inequalities ) and @xcite ( for the last inequality ) to the higher order esfem case .
the high - order version of the inequalities for time independent bilinear forms @xmath181 and @xmath182 were shown in lemma 6.2 in @xcite .
the following results generalises these for the time dependent case .
[ lemma : geometric perturbation errors ] for any @xmath183 , and for their lifts @xmath184 , we have the following bounds @xmath185 where the constants @xmath119 are independent of @xmath40 and @xmath134 , but depend on @xmath133 .
the proof of the first two estimates is a high - order generalization of lemma 5.5 in @xcite , while the proof of the last two estimates is a high - order extension of lemma 7.5 in @xcite .
their proofs follow these references .
both parts are using the geometric estimates showed in lemma [ lemma : geometric est ] . to show the first inequality we estimate as @xmath186 for the second inequality , similarly we have @xmath187 for the third estimate we start by taking the time derivative of the equality @xmath188 , using the first transport property from lemma [ lemma : transport prop ] to obtain @xmath189 hence , using @xmath190 , @xmath191 hence , together with lemma [ lemma : geometric est ] , the bound @xmath192 finishes the proof of the third estimate . for the last inequality we take a similar approach , using the second transport property from lemma [ lemma : transport prop ] and the relation to obtain @xmath193 again , using @xmath190 , together with and the geometric estimates of lemma [ lemma : geometric est ] we obtain @xmath194 completing the proof .
we recall the generalised ritz map for evolving surface pdes from @xcite .
[ def : ritz ] for any given @xmath195 there is a unique @xmath196 such that for all @xmath197 , with the corresponding lift @xmath198 , we have @xmath199 where we let @xmath200 and @xmath201 , to make the forms @xmath202 and @xmath203 positive definite
. then @xmath204 is defined as the lift of @xmath205 , i.e. @xmath206 .
we note here that originally in definition 8.1 in @xcite an extra term appeared involving @xmath207 and the surface velocity , which is not needed for the parabolic case .
the ritz map above is still well - defined .
galerkin orthogonality does not hold in this case , just up to a small defect .
[ lemma : galerkin orthogonality ] for any @xmath195 and @xmath208 @xmath209 where @xmath210 is independent of @xmath211 , @xmath40 and @xmath134 . using the definition of the ritz map and lemma [ lemma : geometric perturbation errors ] we estimate by @xmath212 now we prove higher order error estimates for the ritz map and also for its material derivatives ,
the analogous results for the first order esfem case can be found in section 6 of @xcite , or section 7 of @xcite .
[ theorem : ritz error ] let @xmath213 with @xmath214 for every @xmath5 .
then the error in the ritz map satisfies the bound , for @xmath5 and @xmath215 with sufficiently small @xmath216 , @xmath217 where the constant @xmath119 is independent of @xmath40 and @xmath134 . to ease the presentation
we suppress all time arguments @xmath134 appearing in the norms within the proof ( except some special occasions ) .
\(a ) we first prove the gradient estimate .
starting by the definition of the @xmath218 norm , then using the estimate we have @xmath219 using interpolation error estimate , and for the second term we used the estimate @xmath220 now using young s inequality , and for a sufficiently small ( but @xmath134 independent ) @xmath80 we have the gradient estimate @xmath221 \(b ) the @xmath1-estimate follows from the aubin nitsche trick .
let us consider the problem @xmath222 then by the usual elliptic theory ( see , e.g. section 3.1 of @xcite , or @xcite ) yields : the solution @xmath223 satisfies the bound , with a @xmath119 independent of @xmath134 , @xmath224 by testing the elliptic weak problem with @xmath225 , using again , and using the linear finite element interpolation @xmath143 on @xmath41 , we obtain @xmath226 where for the second term we now used @xmath227 then the combination of the gradient estimate for the ritz map and the interpolation error yields @xmath228 which completes the proof .
since , in general @xmath229 does not hold , we need the following result . [
theorem : ritz mat error ] the error in the material derivatives of the ritz map , for any @xmath230 , satisfies the following bounds , for @xmath5 and @xmath215 with sufficiently small @xmath79 @xmath231 where the constant @xmath160 is independent of @xmath40 and @xmath134 . the proof is a modification of theorem 7.3 in @xcite .
again , to ease the presentation we suppress the @xmath134 argument of the surfaces norms ( except some special occasions ) . for @xmath167 :
( a ) we start by taking the time derivative of the definition of the ritz map , use the transport properties lemma [ lemma : transport prop ] , and apply the definition of the ritz map once more , we arrive at @xmath232 then we obtain @xmath233 where @xmath234 using the geometric estimates of lemma [ lemma : geometric perturbation errors ] , @xmath235 can be estimated as @xmath236 the velocity error estimate lemma [ lemma : velocity estimate ] yields @xmath237 then using @xmath238 as a test function in , and using the error estimates of the ritz map , together with the estimates above , with @xmath80 , we have @xmath239 absorbtion using an @xmath80 , with a sufficiently small @xmath79 , and dividing through yields @xmath240 combining all the previous estimates and using young s inequality , cauchy schwarz inequality , and theorem [ theorem : ritz error ] , for sufficiently small @xmath80 , we obtain @xmath241 then , similarly to the previous proof we have @xmath242 finally the interpolation estimates , young s inequality , absorption using a sufficiently small @xmath80 , yields the gradient estimate .
\(b ) the @xmath1-estimate again follows from the aubin nitsche trick .
let us now consider the problem @xmath243 together with the usual elliptic estimate , for the solution @xmath223 @xmath244 again , @xmath210 is independent of @xmath134 and @xmath40 . following the proof of theorem 6.2 in @xcite , let us first bound @xmath245 where again @xmath143 denotes the linear finite element interpolation operator on @xmath41 .
the pair in the last line can be estimated , using lemma [ lemma : velocity estimate ] , by @xmath246 finally , for the remaining term the proof of theorem 6.2 in @xcite yields @xmath247 for the other bilinear form in we have @xmath248 the combination of all these estimates with , yields@xmath249 by testing the above elliptic weak problem with @xmath225 , and using the above bound and the gradient estimate from ( a ) , we obtain @xmath250 for @xmath251 the proof is analogous .
by combining the error estimates in the ritz map and in its material derivatives and the geometric results of section [ section : geometric estimates ] we prove convergence of the high - order esfem semidiscretisation .
the result is simply shown by repeating the arguments of section 7 in @xcite for our setting , but using the high - order versions of all results : geometric estimates lemma [ lemma : geometric est ] , perturbation estimates of bilinear forms lemma [ lemma : geometric perturbation errors ] and ritz map error estimates lemma [ theorem : ritz error ] and [ theorem : ritz mat error ] .
we follow the approach of section 8.1 of @xcite and section 5 of @xcite . by defining the esfem residual @xmath252 as @xmath253
where @xmath197 , and @xmath254 is the ritz map of the smooth solution @xmath8 .
we now show the optimal order @xmath255-norm estimate of the residual @xmath108 . [
theorem : res bound ] let the solution @xmath8 of the parabolic problem be sufficiently smooth .
then there exist @xmath256 and @xmath79 , such that for all @xmath80 and @xmath17 $ ] , the finite element residual @xmath108 of the ritz map is bounded as @xmath257 \(a ) we start by applying the discrete transport property to the residual equation @xmath258
\(b ) we continue by the transport property with discrete material derivatives from lemma [ lemma : transport prop ] , but for the weak form , with @xmath259 @xmath260 \(c ) subtraction of the two equations , using the definition of the ritz map , and using that@xmath261 holds , we obtain @xmath262 all the pairs can be easily estimated separately as @xmath263 by combining the geometric perturbation estimates of lemma [ lemma : geometric perturbation errors ] , the velocity estimate of lemma [ lemma : velocity estimate ] , and the error estimates of the ritz map from theorem [ theorem : ritz error ] and [ theorem : ritz mat error ] .
the proof is finished using the definition of @xmath255-norm and the equivalence of norms lemma [ lemma : equivalence of discrete norms ] .
using the error estimate for the bdf methods theorem [ theorem : bdf error estimates ] and using the bounds for the semidiscrete residual theorem [ theorem : res bound ] we give here a proof for the fully discrete error estimates of theorem [ theorem : esfem bdf error ] .
the global error is decomposed into two parts @xmath264 and the terms are estimated by results from above .
the first one is estimated by results for bdf methods theorem [ theorem : bdf error estimates ] together with the residual bound theorem [ theorem : res bound ] and by the ritz map error estimates theorem [ theorem : ritz error ] and [ theorem : ritz mat error ] .
the second term is only estimated by the error estimates for the ritz map and for its material derivatives . in remark [ remark : runge
kutta extension ] we noted that the analogous fully discrete results can be shown for algebraically stable implicit runge kutta methods . to be more precise : for the runge kutta analog , instead of theorem [ theorem : bdf error estimates ] one has to use the error bounds of theorem 5.1 in @xcite , but otherwise the proof remains the same .
the implementation of the high - order esfem code follows the typical way of finite element mass matrix , stiffness matrix and load vector assembly , mixed with techniques from isoparametric fem theory .
this was also used for linear esfem , see section 7.2 of @xcite .
similarly to the linear case , a curved element @xmath265 of the @xmath28-order interpolation surface @xmath266 is parametrised over the reference triangle @xmath267 , chosen to be the unit triangle
. then the polynomial map of degree @xmath28 between @xmath265 and @xmath267 is used to compute the local matrices .
all the computations are done on the reference element , using the dunavant quadrature rule , see @xcite . then the local values are summed up to their correct places in the global matrices . in a typical case
the surface is evolved by solving a series of odes , hence only the initial mesh is created based on @xmath29 . naturally , the problem of velocity based grid distortion is still present .
possible overcomes are methods using the deturck trick see @xcite , or using ale finite elements see @xcite , @xcite .
we performed various numerical experiments with quadratic approximation of the surface @xmath16 and using quadratic esfem to illustrate our theoretical results .
let us shortly report on numerical experiments for parabolic problems on a stationary surface , as a benchmark problem .
let @xmath268 be the unit sphere , and let us consider the parabolic surface pde @xmath269 with given initial value and inhomogeneity @xmath270 chosen such , that the solution is @xmath271 .
let @xmath272 and @xmath273 be a series of meshes and time steps , respectively , such that @xmath274 and @xmath275 , with @xmath276 and @xmath277 . for each mesh @xmath278 with corresponding stepsize @xmath279 we numerically solve the surface pde using second order esfem combined with the third order bdf methods .
then , by @xmath280 we denote the error corresponding to the mesh @xmath278 and step size @xmath279 .
we then compute the errors between the lifted numerical solution and the exact solution using the following norm and seminorm : @xmath281 using the above norms , the experimental order of convergence rates ( eocs ) are computed by @xmath282 in table [ table : stationary eocs ] we report on the eocs , for the second order esfem coupled with the bdf3 method , theoretically we expect eoc@xmath283 in the @xmath284 norm , and eoc@xmath285 in the @xmath286 seminorm
. .errors and eocs in the @xmath284 and @xmath286 norms for the stationary problem [ cols=">,>,<,<,<,<",options="header " , ]
the author is grateful for the valuable discussions with prof .
christian lubich during the preparation of the manuscript .
the work of balzs kovcs is funded by deutsche forschungsgemeinschaft , sfb 1173 . | high - order spatial discretisations and full discretisations of parabolic partial differential equations on evolving surfaces are studied .
we prove convergence of the high - order evolving surface finite element method , by showing high - order versions of geometric approximation errors and perturbation error estimates and by the careful error analysis of a modified ritz map .
furthermore , convergence of full discretisations using backward difference formulae and implicit runge kutta methods are also shown .
parabolic problems , evolving surfaces , high - order esfem , ritz map , convergence , bdf and runge kutta methods msc : 35r01 , 65m60 , 65m15 , 65m12 |
in quantum mechanics , it is well known that there is no way to distinguish different nonorthogonal quantum states perfectly without a wrong guess by measurement .
this is because quantum measurement is statistical in nature and it generally destroys the state of the system to be measured .
quantum state discrimination @xcite , as with many ideas in quantum information theory , is most easily understood using the metaphor of a game involving two parties , alice and bob .
alice chooses a state @xmath0 @xmath1 from a set of quantum states @xmath2 with some occurrence probabilities known to both parties .
she gives state @xmath0 to bob , whose task is to identify the given state @xmath0 with one in the set @xmath2 .
when considering such a discrimination problem , two settings are often studied as a standard one . in one
setting , the discrimination success probability is maximized without any restriction on the probability of an incorrect guess @xcite .
this is called minimum - error discrimination since the mean probability of error is minimized as a consequence . on the other hand , a wrong guess is not allowed in the setting of unambiguous discrimination . instead
, the inconclusive result `` i do nt know '' is permitted when the measurement fails to give a definite identification for the input state @xcite .
some other alternative approaches have also been proposed .
one interesting scheme is a maximum - confidence measurement analyzed in refs .
@xcite . in the other scheme ,
considered in refs .
@xcite , the probability of correct discrimination is maximized while the rate of inconclusive results is fixed .
we consider a setting of maximizing the discrimination success probability under the condition that the mean probability of error should not exceed a certain error margin @xmath3 @xcite .
when the error margin @xmath3 is 0 , the setting is equivalent to unambiguous discrimination . in minimum - error discrimination
, no condition is imposed on the probability of error , which means the error margin @xmath3 is 1 .
thus , this formulation naturally unifies the two commonly adopted settings by controlling the error margin . in our previous paper @xcite
, we analyzed discrimination with the error margin between two pure states with general occurrence probabilities and obtained the optimal discrimination success probability in a fully analytic form .
the two - dimensional parameter space consisting of occurrence probabilities and the error margin is divided into three domains : minimum - error , intermediate , and single - state domain . the types of optimal measurement differ depending on the domain . however , for the discrimination problem among more than two pure states , even in the two standard settings it is not easy to obtain analytical solutions , though a great number of works on general theories have been reported @xcite . in this paper , we consider the case of three symmetric states @xcite of a qubit .
we formulate the discrimination problem among three quantum states with general occurrence probabilities in sec . [
sec : formulation ] .
then we analyze the case of three symmetric states of a qubit in secs .
[ sec : pure ] and [ sec : mixed ] .
the optimal measurements are classified into three types by a value of error margin . for an arbitrary error margin
, complete analytical results can be obtained .
we consider the discrimination problem among three quantum states , @xmath4 , @xmath5 , and @xmath6 , with occurrence probabilities @xmath7 , @xmath8 , and @xmath9 , respectively .
measurement is described by a positive operator - valued measure ( povm ) , which consists of four elements @xmath10 .
a measurement outcome labeled @xmath11 = 1 , 2 , or 3 means that the given state is identified with state @xmath0 , and the element @xmath12 produces the inconclusive result .
the joint probability @xmath13 that the given state is @xmath0 and the measurement outcome is @xmath14 is given by @xmath15 the discrimination success probability @xmath16 and the mean probability of error @xmath17 are given by @xmath18 our task is to maximize the discrimination success probability @xmath16 under the conditions [ eq : primal_problem ] @xmath19 where eqs .
( [ eq : povm_positivity ] ) and ( [ eq : povm_completeness_relation ] ) are the usual conditions for a povm , and eq .
( [ eq : error_margin_condition ] ) is the condition that the mean probability of error @xmath17 should not exceed a certain error margin @xmath3 ( @xmath20 ) .
it is clear that unambiguous discrimination is formulated as the case of @xmath21 , while minimum - error discrimination corresponds to the case of @xmath22 .
thus , this scheme continuously interpolates two standard settings of quantum state discrimination .
it is easy to see that this optimization problem can be formulated as one of semidefinite programming ( sdp)@xcite . for applications of sdp to quantum - state discrimination , see refs . @xcite and @xcite .
suppose a hermitian operator @xmath23 and a real number @xmath24 exist such that [ eq : dual_problem ] @xmath25 it is easy to show that @xmath26 gives an upper bound for the discrimination success probability @xmath16 , because @xmath27 } \\ & \ \ \ \ + { \,{\rm tr}\,}{e_{2}\left[y + y\left(\eta_{1}\rho_{1 } + \eta_{3}\rho_{3}\right)\right ] } \\ & \ \ \ \ + { \,{\rm tr}\,}{e_{3}\left[y + y\left(\eta_{1}\rho_{1 } + \eta_{2}\rho_{2}\right)\right ] } \\ & = { \,{\rm tr}\,}{\left(e_{1 } + e_{2 } + e_{3}\right)y } + yp_{\times } \\ & \leq { \,{\rm tr}\,}{y } + ym = d.\end{aligned}\ ] ] it is clear that this upper bound is attained if and only if the following relations hold : @xmath28 = 0 , \label{eq : attainability_condition1 } \\ \nonumber \\ & e_{2}\left[y - \eta_{2}\rho_{2 } + y\left(\eta_{1}\rho_{1 } + \eta_{3}\rho_{3}\right)\right ] = 0 , \label{eq : attainability_condition2 } \\
\nonumber \\ & e_{3}\left[y - \eta_{3}\rho_{3 } + y\left(\eta_{1}\rho_{1 } + \eta_{2}\rho_{2}\right)\right ] = 0 , \label{eq : attainability_condition3 } \\
\nonumber \\ & e_{0}y = 0 , \label{eq : attainability_condition4 } \\
\nonumber \\ & y\left(m - p_{\times}\right ) = 0 .
\label{eq : attainability_condition5}\end{aligned}\ ] ] these five relations are called attainability conditions hereafter . thus , optimal solutions can be obtained if we find a povm @xmath29 , an operator @xmath23 , and a real number @xmath24 which satisfy conditions eqs .
( [ eq : povm_positivity])-([eq : error_margin_condition ] ) , eqs .
( [ eq : y_positivity])-([eq : real_number_y_positivity ] ) , and eqs .
( [ eq : attainability_condition1])-([eq : attainability_condition5 ] ) . minimizing @xmath30 under conditions eqs .
( [ eq : y_positivity])-([eq : real_number_y_positivity ] ) is called the dual problem , whereas the original problem of maximizing @xmath16 under conditions eqs .
( [ eq : povm_positivity])-([eq : error_margin_condition ] ) is referred to as the primal problem .
our strategy to obtain optimal solutions is as follows : we start with the dual problem .
minimization of @xmath30 is sometimes performed by adding extra conditions other than eqs .
( [ eq : y_positivity])-([eq : real_number_y_positivity ] ) on @xmath23 and @xmath24 .
we then construct povm @xmath29 so that eqs .
( [ eq : povm_positivity])-([eq : error_margin_condition ] ) and eqs .
( [ eq : attainability_condition1])-([eq : attainability_condition5 ] ) are fulfilled . in this way
, we obtain the maximum discrimination success probability and the optimal measurement that attains it .
note that the extra conditions in the dual problem do not hamper the strictness of our optimization .
in this section , we consider the case of three symmetric pure states of a qubit @xmath31 , where we assume that the occurrence probabilities @xmath32 are equal and the absolute values of all mutual inner products are the same , @xmath33 .
it is convenient to use the bloch vector representation for @xmath0 and other operators .
the three density operators are given by @xmath34 where @xmath35 are pauli s matrices .
the mutual inner products between the bloch vectors @xmath36 satisfy @xmath37 where @xmath38 is the only one parameter that characterizes the three symmetric pure states of a qubit , and it is in the range of @xmath39 .
since @xmath23 is a 2 @xmath40 2 hermitian operator , it can be written as @xmath41 the conditions given by eqs .
( [ eq : y_positivity])-([eq : y3_positivity ] ) imply that operators @xmath42 , \\ \\
y_{2 } & \equiv y - \frac{1}{3}\left[\rho_{2 } - y\left(\rho_{1 } + \rho_{3}\right)\right ] , \\
\\ y_{3 } & \equiv y - \frac{1}{3}\left[\rho_{3 } - y\left(\rho_{1 } + \rho_{2}\right)\right],\end{aligned}\ ] ] and @xmath23 are positive semidefinite . by using the bloch vector representation , operators @xmath43 , @xmath44 , and @xmath45 are expressed as @xmath46 where we introduced three vectors @xmath47 , @xmath48 , and @xmath49 defined by @xmath50 , \label{eq : vector_a1 } \\
\nonumber \\ { \mbox{\boldmath $ a$}}_{2 } & \equiv \frac{1}{3}\left[{\mbox{\boldmath $ n$}}_{2 } - y\left({\mbox{\boldmath $ n$}}_{1 } + { \mbox{\boldmath $ n$}}_{3}\right)\right ] , \label{eq : vector_a2 } \\
\nonumber \\ { \mbox{\boldmath $ a$}}_{3 } & \equiv \frac{1}{3}\left[{\mbox{\boldmath $ n$}}_{3 } - y\left({\mbox{\boldmath $ n$}}_{1 } + { \mbox{\boldmath $ n$}}_{2}\right)\right ] .
\label{eq : vector_a3}\end{aligned}\ ] ] since the smaller eigenvalue of operators @xmath43 , @xmath44 , @xmath45 , and @xmath23 are positive , we obtain the following four inequalities for @xmath51 and @xmath52 : @xmath53 moreover , the upper bound @xmath30 can be rewritten as @xmath54 in terms of parameters @xmath55 , the dual problem takes the following form : minimize [ eq : formulation_of_alpha_and_beta ] @xmath56 subject to @xmath57 in what follows , we present the main results first , leaving their derivation to subsequent subsections . the parameter space is divided into the following three domains , @xmath58 where two critical error margins @xmath59 and @xmath60 are defined by @xmath61 the maximum discrimination success probability in each domain
is found to be @xmath62 where @xmath63 is a positive constant defined by @xmath64 suppose the error margin is so large that the constraint on the probability of error is inactive , which means that the probability of error @xmath17 in the optimal measurement is strictly smaller than the error margin @xmath3 .
then , the optimal measurement is that of minimum - error discrimination .
hereafter , the domain where this is the case is called minimum - error domain . in the minimum - error domain , the condition given by eq .
( [ eq : attainability_condition5 ] ) implies that parameter @xmath24 should be 0 since @xmath65 . from this consideration
, we can rewrite the discrimination problem as follows : minimize @xmath66 subject to @xmath67 the condition given by eq .
( [ eq : y_eigenvalue_positivity ] ) can be omitted , since eq .
( [ eq : y_eigenvalue_positivity ] ) is a consequence of the conditions eqs .
( [ eq : y1_eigenvalue_positivity_me])-([eq : y3_eigenvalue_positivity_me ] ) .
this is obvious from the inequalities @xmath68 which are obtained by applying the triangle inequality on the right - hand side of eqs .
( [ eq : y1_eigenvalue_positivity_me])-([eq : y3_eigenvalue_positivity_me ] ) .
we construct a solution where all povm elements @xmath69 , @xmath70 , and @xmath71 are of rank 1 .
thus , we assume that equality holds in eqs .
( [ eq : y1_eigenvalue_positivity_me])-([eq : y3_eigenvalue_positivity_me ] ) as follows : @xmath72 by solving eqs .
( [ eq : alpha_beta_equation1_me])-([eq : alpha_beta_equation3_me ] ) , it turns out that vector @xmath52 is given by @xmath73 and @xmath51 is given by @xmath74 we construct the set of povm @xmath75 which attains the upper bound @xmath30 given by eq .
( [ eq : upper_bound_pure_me ] ) .
the element @xmath76 for the inconclusive result is taken to be 0 in this domain and the attainability condition eq .
( [ eq : attainability_condition4 ] ) is satisfied .
the attainability conditions given by eqs .
( [ eq : attainability_condition1])-([eq : attainability_condition3 ] ) require that povm elements @xmath69 , @xmath70 , and @xmath71 take the following form : @xmath77 here , we defined three unit vectors , @xmath78 where vector @xmath52 is given in eq .
( [ eq : beta_pure_me ] ) .
we determine the coefficients @xmath79 @xmath80 so that the conditions given by eqs .
( [ eq : povm_positivity])-([eq : error_margin_condition ] ) are satisfied . from the completeness relation of povm given by eq .
( [ eq : povm_completeness_relation ] ) , we obtain the two relations for the coefficients @xmath79 as follows : @xmath81 in eq .
( [ eq : part_of_vector_me ] ) , each of the three vectors @xmath82 , @xmath83 , and @xmath84 is expressed by the three bloch vectors @xmath85 , @xmath86 , and @xmath87 . from the linear independence of the three bloch vectors @xmath85 , @xmath86 , and @xmath87 , we find @xmath88 . therefore , the coefficient @xmath79 @xmath80 is given by @xmath89 it is clear that the set of povm @xmath90 satisfies eq .
( [ eq : povm_positivity ] ) since @xmath91 . by the remaining condition eq .
( [ eq : error_margin_condition ] ) , we find @xmath92 where the mean probability of error @xmath17 is calculated by using the povm constructed above .
thus , if the error margin @xmath3 is in the range @xmath93 , the upper bound of eq .
( [ eq : upper_bound_pure_me ] ) is attained and the maximum discrimination success probability is given by @xmath94 in this subsection , we construct a solution where all povm elements @xmath69 , @xmath70 , @xmath71 , and @xmath12 are nonzero .
the attainability conditions given by eqs .
( [ eq : attainability_condition1])-([eq : attainability_condition4 ] ) imply that operators @xmath43 , @xmath44 , @xmath45 , and @xmath23 are rank 1 at most .
that is , the smaller eigenvalues of operators @xmath43 , @xmath44 , @xmath45 , and @xmath23are all 0 .
thus , we assume that equality holds in eqs .
( [ eq : y1_eigenvalue_positivity])-([eq : y_eigenvalue_positivity ] ) as follows : @xmath95 solving eqs .
( [ eq : alpha_beta_equation1])-([eq : alpha_beta_equation4 ] ) requires a rather complicated and long calculation .
it turns out that parameter @xmath24 must satisfy @xmath96 , vector @xmath52 is given by @xmath97 and @xmath51 is given by @xmath98 therefore , we obtain an upper bound for the discrimination success probability as a function of parameter @xmath24 : @xmath99 as for the double signs in the above equation , we take a negative one to obtain a smaller upper bound . correspondingly , a negative sign is also taken in the double signs of eqs .
( [ eq : vector_beta_pure ] ) and ( [ eq : alpha_pure ] ) hereafter
. now we determine parameter @xmath24 so that the upper bound @xmath30 is minimized .
let us begin by looking at how the upper bound @xmath30 changes in the range of parameter @xmath96 . by differentiating the upper bound @xmath30 with respect to parameter @xmath24
, we have @xmath100 a positive constant @xmath63 is defined by eq .
( [ eq : a ] ) .
if the error margin is @xmath101 , it turns out that the upper bound @xmath30 is a monotone decreasing function with respect to parameter @xmath24 . in this case , the upper bound @xmath30 takes the minimum @xmath102 at @xmath103 .
on the other hand , in the case of @xmath104 , the quadratic equation for a parameter @xmath24 produced by @xmath105 has the roots @xmath106 given by @xmath107 as for the double signs of the roots @xmath106 , we take a positive one since @xmath96 . here , we consider the two cases .
one is the case of @xmath108 , which can be rewritten as a range of error margin @xmath3 , @xmath109 where @xmath60 on the right - hand side is defined in eq .
( [ eq : mc_pure ] ) .
this domain is called the linear domain . in linear domain ,
the upper bound @xmath30 is minimum at @xmath103 .
moreover , @xmath110 since @xmath52 and @xmath51 become 0 .
therefore , the upper bound @xmath30 in linear domain is given by @xmath111 which is linear with respect to the error margin @xmath3 .
the other is the case of @xmath112 , where we have @xmath113 this domain is called the intermediate domain . in the intermediate domain , the upper bound @xmath30 is minimum at @xmath114 . using @xmath23 and @xmath24 , which minimize an upper bound for the discrimination success probability
, we obtain the minimum of the upper bound @xmath115 to be @xmath116 in the following subsections , we construct the optimal povm to achieve the obtained upper bound of each domain . in linear domain , we have @xmath110 since @xmath117 .
this shows that the attainability condition given by eq .
( [ eq : attainability_condition4 ] ) does not give any restriction on povm element @xmath12 .
the attainability conditions given by eqs .
( [ eq : attainability_condition1])-([eq : attainability_condition3 ] ) require that povm elements @xmath69 , @xmath70 , and @xmath71 take the following form : @xmath118 here , we assumed that the coefficient @xmath119 does not depend on index @xmath11 @xmath80 and the vectors @xmath120 are defined by @xmath121 where @xmath122 are given in eqs .
( [ eq : vector_a1])-([eq : vector_a3 ] ) with @xmath103 .
the question is whether the coefficient @xmath119 can be chosen so that the set @xmath90 satisfies the conditions given by eqs .
( [ eq : povm_positivity])-([eq : error_margin_condition ] ) , and ( [ eq : attainability_condition5 ] ) . the conditions given in eqs .
( [ eq : error_margin_condition ] ) and ( [ eq : attainability_condition5 ] ) are reduced to @xmath123 since @xmath103 . calculating @xmath17 by the povm constructed above
, we find that the coefficient @xmath119 is given by @xmath124 and @xmath119 is positive since @xmath39 .
this shows that povm elements @xmath69 , @xmath70 , and @xmath71 satisfy eq .
( [ eq : povm_positivity ] ) .
in addition , povm element @xmath12 is obtained from the completeness relation of povm given in eq .
( [ eq : povm_completeness_relation ] ) as follows : @xmath125 here , the smaller eigenvalue of povm element @xmath12 , @xmath126 should be positive since @xmath127 .
this is satisfied since the error margin @xmath3 is in the range @xmath128 .
therefore , if the error margin @xmath3 is in the range @xmath128 , the upper bound of eq .
( [ eq : upper_bound_pure_ld ] ) is attained and the maximum discrimination success probability is given by @xmath129 we construct the set of povm @xmath10 which attains the upper bound @xmath30 given by eq .
( [ eq : upper_bound_pure_id ] ) .
the attainability conditions given by eq .
( [ eq : attainability_condition1])-([eq : attainability_condition4 ] ) require that povm elements @xmath69 , @xmath70 , @xmath71 , and @xmath12 take the form @xmath130 where we defined @xmath131 here , @xmath52 is given in eq .
( [ eq : vector_beta_pure ] ) with @xmath106 given by eq .
( [ eq : ye_pure ] ) substituted for @xmath24 , and vectors @xmath122 @xmath80 are given in eqs .
( [ eq : vector_a1 ] ) , ( [ eq : vector_a2 ] ) , and ( [ eq : vector_a3 ] ) with @xmath114 . the completeness relation of povm given by eq . ( [ eq : povm_completeness_relation ] )
is now expressed as @xmath132 in eq .
( [ eq : part_of_vector_id ] ) , each of the four vectors @xmath82 , @xmath83 , @xmath84 , and @xmath133 is expressed by the three bloch vectors @xmath85 , @xmath86 , and @xmath87 . from the linear independence of the three bloch vectors @xmath85 , @xmath86 , and @xmath87 , we can see that there is the relation @xmath88 .
hereafter , we write @xmath119 for @xmath79 . after a rather long calculation , we obtain @xmath134 , \\ \\
c_{0 } & = 2 - 3c.\end{aligned}\ ] ] the coefficients @xmath119 and @xmath135 should be positive to satisfy eq .
( [ eq : povm_positivity ] ) .
this is satisfied since the error margin @xmath3 is in the range @xmath136 .
the remaining conditions are eqs .
( [ eq : error_margin_condition ] ) and ( [ eq : attainability_condition5 ] ) , which are reduced to @xmath123 since @xmath96 .
we can explicitly verify that the relation @xmath123 holds after a long calculation by using the povm constructed above .
this is not a coincidence , but a consequence of how we determined parameter @xmath24 .
parameter @xmath24 was determined so that the upper bound @xmath30 given by eq .
( [ eq : upper_bound_d_pure ] ) is minimized : @xmath137 using the same argument given in ref .
@xcite , we can show that @xmath138 , which means that minimization of @xmath30 leads to the relation @xmath123 .
thus , if the error margin @xmath3 is in the range @xmath136 , the upper bound of eq .
( [ eq : upper_bound_pure_id ] ) is attained and the maximum discrimination success probability is given by @xmath139 we consider the set of states defined by the following bloch vectors @xmath85 , @xmath86 , and @xmath87 : @xmath140 the mutual inner products between bloch vectors is @xmath141 @xmath142 .
[ solid ( red ) line ] and probability of an inconclusive result @xmath143 [ dashed ( green ) line ] vs error margin @xmath3.,width=264 ] figure [ fig : example ] displays the maximum discrimination success probability @xmath144 and the probability of an inconclusive result @xmath143 as a function of the error margin @xmath3 . as the error margin increases from 0 to 1 , the type of optimal measurement varies in the following way : from the linear to the intermediate type at @xmath145 , and the minimum - error type at @xmath146 .
the maximum discrimination success probability @xmath144 and the probability of an inconclusive result @xmath143 do not depend on the error margin @xmath3 in the range @xmath93 .
moreover , the curves of @xmath144 and @xmath143 clearly show the border between the minimum - error and the intermediate domain , though the curves are smooth at @xmath145 . in our previous paper @xcite , we analyzed discrimination with the error margin between two pure states with general occurrence probabilities and found that there is a domain where omitting one of the states to be discriminated is optimal if the error margin is sufficiently small .
however , there is no such domain for the symmetric states considered in the present paper , where there is a symmetry with respect to interchange of the states .
in this section , we consider that three states to be discriminated , @xmath4 , @xmath5 , and @xmath6 , are mixed and symmetric and satisfy the following conditions : @xmath147 the mutual inner products between the bloch vectors are parametrized as @xmath148 where @xmath149 and @xmath150 .
the symmetric set of three mixed states of a qubit is characterized by two parameters , @xmath151 and @xmath38 .
the problem is formulated in exactly the same way as in eq .
( [ eq : formulation_of_alpha_and_beta ] ) . the optimal solution can be obtained as in the case of pure states , though the calculation becomes much more complex . in
what follows , we present the results of the case of three symmetric mixed states . to make the expressions simpler ,
we define @xmath152 we obtain the maximum discrimination success probability in the case of three symmetric mixed states of a qubit as @xmath153 where @xmath63 is given by @xmath154 and @xmath59 and @xmath60 are defined by @xmath155 note that , when @xmath156 , this reproduces the results obtained in sec .
[ sec : pure ] .
we have considered a state discrimination problem which interpolates minimum - error and unambiguous discriminations by introducing a margin for the probability of error . in the case of three symmetric pure states
, we obtained the optimal success probability in a fully analytic from .
we also showed that our method can be applied to the three symmetric mixed states .
the optimal measurement is classified into three types .
one of the three types of measurement is optimal , depending on the error margin .
we note that this classification is also done by the rank of povm element @xmath12 : @xmath157 when rank@xmath158 , the optimal measurement is that of minimum - error domain . however , physical implication of the difference between the case of rank@xmath159 and the case of rank@xmath160 is not very clear .
_ note added in proof . _
recently we became aware of two recent related works @xcite .
in both works , discrimination with a fixed rate of inconclusive results is applied to the set of symmetric states and it is argued that the results can be transformed to solutions for discrimination with error margin . | we consider a state discrimination problem which deals with settings of minimum - error and unambiguous discrimination systematically by introducing a margin for the probability of an incorrect guess .
we analyze discrimination of three symmetric pure states of a qubit .
the measurements are classified into three types , and one of the three types is optimal depending on the value of the error margin . the problem is formulated as one of semidefinite programming . starting with the dual problem derived from the primal one , we analytically obtain the optimal success probability and the optimal measurement that attains it in each domain of the error margin .
moreover , we analyze the case of three symmetric mixed states of a qubit . |
in the past years , the tachyonic lump solutions @xcite as well as the tachyon potential @xcite have been worked out in bosonic cubic string field theory @xcite using the level truncation scheme . however , these calculations were carried out only in the feynman - siegel gauge . although its validity has been checked in @xcite , it would be interesting to see whether other gauge choices are possible .
in fact , it has recently been found in @xcite that the feynman - siegel gauge condition has a finite range of validity , and that some other gauge choices lead to similar tachyon potentials to that in the feynman - siegel gauge .
it is the problem of choosing the gauge fixing condition(s ) that we will address in the first half of this paper .
we will see that we can successfully construct lump solutions with nearly correct tensions in several gauges , and that the widths of the lumps are independent of these gauge chioces at least at level ( 2,4 ) .
concerning this point , in the recent studies of so - called ` vacuum string field theory'@xcite it was guessed that the width of the lump solution was a gauge - dependent quantity @xcite . in more detail , let us consider the center - of - mass coordinate @xmath1 of the string and its conjugate momentum @xmath2 , obeying the canonical commutation relation @xmath3=i\eta^{\mu\nu}. \label{eq : a}\ ] ] combining these two operators , we can move to the oscillator representation @xcite of the zero modes as @xcite @xmath4 it is easily found that they satisfy the commutation relation @xmath5= \eta^{\mu\nu}$ ] for the creation - annihilation operators irrespective of the value of the newly introduced parameter @xmath6 .
moreover , it has been observed in @xcite that the value of the ratio of the lump tensions seems to converge to the expected single value for all @xmath6 , while the extent of the lump in spacetime clearly depends on @xmath6 . from these facts , the authors of @xcite
has proposed that the solutions of different widths are related to each other by the gauge transformation which does not affect the ghost part of the string field . though our result suggests that the size of the lump can not be changed by gauge transformation in the ` ordinary ' cubic string field theory , in the flat spacetime , as opposed to the vacuum string field theory . ]
it does not immediately contradict the above proposal because the relation between vacuum string field theory and the ordinary cubic string field theory is yet to be clarified .
since some of the symmetries in the closed string vacuum are broken in a d - brane background , it is conceivable that the ordinary cubic string field theory we will use has fewer gauge degrees of freedom than the vacuum string field theory .
if the gauge symmetry which actually changes the width of the lump is broken in the presence of a d - brane background , we can not see such an effect using the ordinary cubic open string field theory on a d - brane . in the second half of this paper
, we consider the fluctuation spectrum around the codimension-1 lump solution constructed in the modified level truncation scheme @xcite . by now , it has been verified that the spacetime - independent non - perturbative vacuum solution in the ordinary cubic string field theory does not support any perturbative open string excitations @xcite within the level truncation scheme . on the other hand , if the conjecture that the tachyonic lump solutions represent the lower - dimensional d - branes @xcite is true , the fluctuation spectrum around the lump solution should agree with that of the corresponding d - brane .
analyses of this kind have so far been done in the tachyonic @xmath7 scalar field theory model - adic string theory @xcite and in field theory models for tachyon dynamics @xcite . ]
@xcite , the result being that the spectrum on the lump contains a tachyon with @xmath8 , a massless scalar and an approximately massless gauge field ( it is reviewed in section 4.1 of @xcite ) .
at this level of approximation , the mass of the tachyon is not so close to the expected value @xmath9 and the tension of the lump is only 78% of the correct answer .
we will examine the fluctuation spectrum on the lump solution to a higher degree of accuracy by taking into account the scalar component fields at level 2 , and find that there is a tachyon with mass @xmath10 .
this paper is organized as follows . in section 2
we discuss the dependence of the tachyonic lump on the gauge choice at level ( 2,4 ) approximation . in section 3 , we determine the low - lying mass spectrum around the lump solution in the feynman - siegel gauge , and find that a tachyon appears with the nearly correct mass squared .
section 4 includes a summary of our results and discussions on some points .
numerical data for expectation values and quite lengthy expressions for the action are collected in appendices .
in this section we study the dependence of the lump solutions on the gauge choice .
we consider the possibility of imposing gauge fixing conditions other than the feynman - siegel gauge @xmath11 on the string field as well as that of leaving the gauge unfixed . in the modified level truncation scheme introduced in @xcite , momentum in the @xmath12-direction , in which we will construct non - trivial field configurations ,
is discretized as @xmath13 with @xmath14 integer by compactifying the @xmath12-direction on a circle of radius @xmath15 .
this compact momentum is taken to contribute to the level as @xmath16 where @xmath17 denotes the conformal weight of the state excluding the contribution from the momentum factor .
( for more details , see @xcite and chapter 4 of @xcite . ) to this end , we consider the level ( 2,4 ) truncation approximation . at this level ,
the string field without gauge fixing , restricted to the twist - even sector , is expanded as @xmath18 where @xmath19 and @xmath20 is the @xmath21-invariant vacuum .
we have denoted by @xmath22 the 25-dimensional manifold excluding @xmath12 , and by @xmath23 the virasoro generators of the conformal field theory of central charge 25 associated with @xmath22 . in finding the lump solution which is symmetric under the reflection
@xmath24 , we can put the following constraints on @xmath25 s , @xmath26 so that @xmath27 .
the range of @xmath14 over which the summation in eq.([eq : d ] ) is taken depends on the value of @xmath15 . in this section our choices of @xmath15 are @xmath28 and @xmath29 , in which cases @xmath30 at level 2 . due to the reality condition @xmath31 on the string field , the component fields @xmath32 must be real . substituting the level - expanded string field ( [ eq : d ] ) into the cubic action @xmath33 on a d@xmath34-brane , we have found the level ( 2,4 ) truncated gauge - unfixed action to be @xmath35 , \label{eq : h}\end{aligned}\ ] ] where @xmath36 , @xmath37 , and note that our normalization convention is such that every component field is dimensionless and that we are not setting @xmath38 to any fixed value . in spite of leaving the gauge degrees of freedom unfixed
, the procedure of level truncation actually breaks the gauge invariance , so that the equations of motion obtained by varying the action ( [ eq : h ] ) have a discrete set of solutions , instead of a continuous family .
hence we can find the _ isolated _ ` closed string vacuum ' solution and the lump solutions for two values ( [ eq : f ] ) of radius by solving the simultaneous equations of motion numerically , even without any gauge fixing .
the expectation values found this way are given in table [ tab : taa ] in appendix [ sec : appa ] . putting these values back to the above action
, we can obtain the numerical values for the depth of the tachyon potential and for the tensions of the lump solutions using the following formulae @xcite , respectively , @xmath39 where we have denoted by @xmath40 the tension of the original d@xmath34-brane , by @xmath41 the closed string vacuum configuration of the string field , and by @xmath42 the lump configuration .
we have used the relation @xcite @xmath43 between the d@xmath34-brane tension and the open string coupling on the d@xmath34-brane in the most right hand side of eq.([eq : i ] ) .
the conjectured values for @xmath44 and @xmath45 are 1 .
our results obtained with the gauge left unfixed are @xmath46 these values are summarized in table [ tab : ta ] , together with the gauge - fixed ones to be explained below .
next we consider fixing the gauge .
as we mentioned above , the gauge invariance has actually been broken by the level truncation . however , this breakdown is not under control so that the gauge - fixing is needed to obtain more definite results .
the first gauge fixing condition we try to choose is the feynman - siegel gauge @xmath47 whose validity at the linearized level was established in @xcite , and its range of validity in the non - perturbative case has been argued in @xcite . in terms of the component fields appearing in eq.([eq : d ] ) , this condition is restated simply as @xmath48 as alternatives we consider the following three choices @xmath49 whose component forms are , respectively , @xmath50 as suggested in @xcite . the validity of these gauges at the linearized level can be verified in a similar way to the case of the feynman - siegel gauge , at least in the level ( 2,4 ) truncation scheme . to give an example ,
let us look at the @xmath51-gauge closely .
suppose that the initial string field @xmath52 is not annihilated by @xmath51 .
acting on @xmath52 with the linearized gauge transformation , we reach the transformed field @xmath53 where the gauge parameter @xmath54 must have ghost number 0 .
if @xmath54 is taken to satisfy @xmath55 then it follows that @xmath56 so we have found that there is a representative obeying the gauge fixing condition @xmath57 in each gauge equivalent class , _ if there exists a suitable gauge parameter @xmath54 satisfying ( [ eq : p])_. the string field @xmath54 of ghost number 0 is expanded as @xmath58 since we are focusing on the even level fields thanks to the twist symmetry , the first term ( at oscillator level 1 ) can be dropped .
in addition , the third and the still higher terms do not contribute at level 2 .
therefore , we have only to consider the 1-parameter gauge degree of freedom @xmath59 at this level of approximation . it is easily found that the above @xmath54 satisfies both conditions in ( [ eq : p ] ) if we take @xmath60 to be equal to @xmath61 .
furthermore , it is obvious that there is no residual gauge degree of freedom after fixing the value of @xmath60 .
similarly , we can reach the @xmath62- and @xmath63-gauge by taking @xmath64 and @xmath65 , respectively , with no residual gauge degrees of freedom left unfixed .
putting aside the problem on the validity of these gauge fixing conditions at higher levels , we will try to construct lump solutions as well as the tachyon potential using these gauge choices . and we will test one more gauge fixing condition @xmath66 which was proposed in @xcite as a modified version of the feynman - siegel gauge .
this modification is necessary in vacuum string field theory with a non - standard brst operator @xmath67 made purely out of ghost operators because of the problems described in @xcite .
though the ordinary cubic string field theory has no problem about the feynman - siegel gauge , it is worth trying it in our context .
written in terms of the component fields , @xmath68 gives @xmath69 after putting the gauge fixing conditions ( [ eq : t]),([eq : n ] ) or ( [ eq : s ] ) on the action ( [ eq : h ] ) , we look for solutions by extremizing the resulting action with respect to the remaining field variables .
the field values which solve the equations of motion are shown in table [ tab : taa][tab : tac ] in appendix [ sec : appa ] , and the values of the lump tensions are displayed in table [ tab : ta ] .
.the numerical values of the depth @xmath70 of the tachyon potential and those of the lump tensions @xmath71 for various gauge choices , at level ( 2,4 ) approximation . [ cols="^,^,^,^ " , ]
we write below the explicit expression of the action obtained by substituting the string field ( [ eq : ac ] ) into the cubic action ( [ eq : g ] ) . @xmath72 \\ & & + \int d^{25}\!k_1d^{25}\!k_2d^{25}\!k_3\ \delta^{25}\left(\sum_{i=1}^3k_i\right ) k^3\biggl[\frac{1}{3 } \sum_{n_1,n_2,n_3=-2}^2\delta_{n_1+n_2+n_3}\widetilde{\phi}_{n_1}(k_1 ) \widetilde{\phi}_{n_2}(k_2)\widetilde{\phi}_{n_3}(k_3 ) \\ & & + \frac{11}{27}\sum_{n=-1}^1\widetilde{\phi}_n(k_1)\widetilde{\phi}_{-n}(k_2)\widetilde{u}(k_3 ) -\frac{2}{9}\sqrt{2{\alpha^{\prime}}}\sum_{n=-1}^1\widetilde{\phi}_n(k_1)ik_2^{\mu}\widetilde{\check{b}}_{\mu } ( k_2)\widetilde{\phi}_{-n}(k_3 ) \\ & & + \frac{1}{27}\sum_{n=-1}^1\widetilde{\phi}_n(k_1)\widetilde{\phi}_{-n}(k_3)\biggl\ { 8{\alpha^{\prime}}(k_1-k_3)^{\mu}(k_1-k_3)^{\nu}\widetilde{b}_{\mu\nu}(k_2)+\frac{32{\alpha^{\prime}}}{r}n(k_1-k_3)^{\mu } \widetilde{b}_{\mu x}(k_2 ) \\ & & + \frac{32{\alpha^{\prime}}}{r^2}n^2\widetilde{b}_{xx}(k_2)-5\widetilde{b}_{xx}(k_2)-5 \eta^{\mu\nu}_{(25 ) } \widetilde{b}_{\mu\nu}(k_2)\biggr\ } \\ & & + \frac{19}{243}\widetilde{\phi}_0(k_1)\widetilde{u}(k_2)\widetilde{u}(k_3)+\frac{8{\alpha^{\prime}}}{81 } \widetilde{\phi}_0(k_1)ik_2^{\mu}\widetilde{\check{b}}_{\mu}(k_2)ik_3^{\nu } \widetilde{\check{b}}_{\nu}(k_3 ) \\ & & -\frac{256}{243}\widetilde{\phi}_0(k_1)\left(i\widetilde{\check{b}}_x(k_2)i \widetilde{\check{b}}_x(k_3)+\eta^{\mu\nu}_{(25)}i\widetilde{\check{b}}_{\mu}(k_2)i \widetilde{\check{b}}_{\nu}(k_3)\right ) \\ & & + \frac{1}{729}\widetilde{\phi}_0(k_1)\left\ { 8{\alpha^{\prime}}(k_2-k_1)^{\rho}(k_2-k_1)^{\sigma } \widetilde{b}_{\rho\sigma}(k_3)-5\widetilde{b}_{xx}(k_3)-5\eta^{\rho\sigma}_{(25)}\widetilde{b}_{\rho \sigma}(k_3)\right\ } \\ & & \quad \times\left\ { 8{\alpha^{\prime}}(k_1-k_3)^{\mu}(k_1-k_3)^{\nu } \widetilde{b}_{\mu\nu}(k_2)-5\widetilde{b}_{xx}(k_2)-5\eta^{\mu\nu}_{(25)}\widetilde{b}_{\mu \nu}(k_2)\right\ } \\ & & + \frac{512}{729}\widetilde{\phi}_0(k_1)\left\ { \widetilde{b}_{xx}(k_2)\widetilde{b}_{xx}(k_3 ) + \widetilde{b}_{\mu\nu}(k_2)\widetilde{b}^{\mu\nu}(k_3)+2\eta^{\mu\rho}_{(25 ) } \widetilde{b}_{\mu x}(k_2)\widetilde{b}_{\rho x}(k_3 ) \right\ } \\ & & + \frac{512}{729}{\alpha^{\prime}}\widetilde{\phi}_0(k_1)(k_1-k_3)^{\mu}(k_2-k_1)^{\sigma}\left\ { \widetilde{b}_{\mu x}(k_2)\widetilde{b}_{\sigma x}(k_3)+\eta^{\nu\rho}_{(25 ) } \widetilde{b}_{\mu\nu}(k_2)\widetilde{b}_{\rho\sigma}(k_3)\right\ } \\ & & + \frac{22}{729}\left\ { 8{\alpha^{\prime}}(k_2-k_1)^{\mu}(k_2-k_1)^{\nu}\widetilde{b}_{\mu\nu}(k_3 ) -5\widetilde{b}_{xx}(k_3)-5\eta^{\mu\nu}_{(25)}\widetilde{b}_{\mu\nu}(k_3)\right\ } \widetilde{\phi}_0(k_1)\widetilde{u}(k_2 ) \\ & & -\frac{44\sqrt{2{\alpha^{\prime}}}}{243}\widetilde{\phi}_0(k_1)\widetilde{u}(k_2)ik_3^{\mu } \widetilde{\check{b}}_{\mu}(k_3)+\frac{512}{729}\sqrt{2{\alpha^{\prime}}}\widetilde{\phi}_0(k_1)\biggl\ { ( k_2-k_1)^{\rho}i\widetilde{\check{b}}_x(k_2)\widetilde{b}_{\rho x}(k_3 ) \\ & & \quad + ( k_2-k_1)^{\rho}\eta^{\mu\nu}i\widetilde{\check{b}}_{\mu}(k_2)\widetilde{b}_{\nu\rho } ( k_3)\biggr\ } -\frac{4\sqrt{2{\alpha^{\prime}}}}{243}\widetilde{\phi}_0(k_1)ik_2^{\mu } \widetilde{\check{b}}_{\mu}(k_2 ) \\ & & \quad \times \biggl\ { 8{\alpha^{\prime}}(k_2-k_1)^{\nu}(k_2-k_1)^{\rho}\widetilde{b}_{\nu\rho}(k_3)-5\widetilde{b}_{xx } ( k_3)-5\eta^{\nu\rho}_{(25)}\widetilde{b}_{\nu\rho}(k_3)\biggr\}\biggr],\end{aligned}\ ] ] where we have defined the tilded fields as @xmath73 and so on .
we show the explicit form of the matrix @xmath75 defined in eq.([eq : an ] ) .
the assignment of indices follows from @xmath76 in eq.([eq : am ] ) . for simplicity , we omit the overlines on the understanding that all the field variables appearing below represent their expectation values .
@xmath77k^{3 - 2{\alpha^{\prime}}k^2-\frac{2{\alpha^{\prime}}}{r^2 } } , \\
m_{33}&=&\frac{1}{2}\left({\alpha^{\prime}}k^2 - 1\right)+\phi_0 k^{3 - 2{\alpha^{\prime}}k^2}+ \frac{11}{27}u k^{3 - 2{\alpha^{\prime}}k^2 } \\ & & + \frac{1}{27}\left[\left(16{\alpha^{\prime}}k^2-\frac{125}{2 } \right)w-\frac{5}{2}v\right]k^{3 - 2{\alpha^{\prime}}k^2 } , \\
m_{44}&=&\frac{1}{2}\left({\alpha^{\prime}}k^2+\frac{{\alpha^{\prime}}}{r^2}-1\right)+\phi_0 k^{3 - 2{\alpha^{\prime}}k^2-\frac{2{\alpha^{\prime}}}{r^2}}+ \frac{11}{27}u k^{3 - 2{\alpha^{\prime}}k^2-\frac{2{\alpha^{\prime}}}{r^2 } } \\ & & + \frac{1}{27}\left[\left(16{\alpha^{\prime}}k^2-\frac{125}{2 } \right)w+\left(\frac{16{\alpha^{\prime}}}{r^2}-\frac{5}{2}\right)v\right]k^{3 - 2{\alpha^{\prime}}k^2-\frac{2{\alpha^{\prime}}}{r^2 } } , \\
m_{55}&=&\frac{1}{2}\left({\alpha^{\prime}}k^2+\frac{4{\alpha^{\prime}}}{r^2}-1\right)+\phi_0 k^{3 - 2{\alpha^{\prime}}k^2-\frac{8{\alpha^{\prime}}}{r^2 } } , \\
m_{12}&=&\phi_1k^{3 - 2{\alpha^{\prime}}k^2-\frac{6{\alpha^{\prime}}}{r^2 } } , \\
m_{13}&=&\phi_2k^{3 - 2{\alpha^{\prime}}k^2-\frac{8{\alpha^{\prime}}}{r^2 } } , \\
m_{14}&=&m_{15}=m_{25}=m_{41}=m_{51}=m_{52}=0 , \\
m_{21}&=&\phi_{-1}k^{3 - 2{\alpha^{\prime}}k^2-\frac{6{\alpha^{\prime}}}{r^2 } } , \\
m_{23}&=&\phi_1k^{3 - 2{\alpha^{\prime}}k^2-\frac{2{\alpha^{\prime}}}{r^2 } } , \\
m_{24}&=&\phi_2k^{3 - 2{\alpha^{\prime}}k^2-\frac{6{\alpha^{\prime}}}{r^2 } } , \\
m_{31}&=&\phi_{-2}k^{3 - 2{\alpha^{\prime}}k^2-\frac{8{\alpha^{\prime}}}{r^2 } } , \\
m_{32}&=&\phi_{-1}k^{3 - 2{\alpha^{\prime}}k^2-\frac{2{\alpha^{\prime}}}{r^2 } } , \\
m_{34}&=&\phi_1k^{3 - 2{\alpha^{\prime}}k^2-\frac{2{\alpha^{\prime}}}{r^2 } } , \\
m_{35}&=&\phi_2k^{3 - 2{\alpha^{\prime}}k^2-\frac{8{\alpha^{\prime}}}{r^2 } } , \\
m_{42}&=&\phi_{-2}k^{3 - 2{\alpha^{\prime}}k^2-\frac{6{\alpha^{\prime}}}{r^2 } } , \\
m_{43}&=&\phi_{-1}k^{3 - 2{\alpha^{\prime}}k^2-\frac{2{\alpha^{\prime}}}{r^2 } } , \\
m_{45}&=&\phi_1k^{3 - 2{\alpha^{\prime}}k^2-\frac{6{\alpha^{\prime}}}{r^2 } } , \\
m_{53}&=&\phi_{-2}k^{3 - 2{\alpha^{\prime}}k^2-\frac{8{\alpha^{\prime}}}{r^2 } } , \\
m_{54}&=&\phi_{-1}k^{3 - 2{\alpha^{\prime}}k^2-\frac{6{\alpha^{\prime}}}{r^2 } } , \\ m_{66}&=&-\frac{1}{2}({\alpha^{\prime}}k^2 + 1)+\frac{19}{243}\phi_0k^{3 - 2{\alpha^{\prime}}k^2 } , \\
m_{67}&=&m_{76}=m_{6,10}=m_{10,6}=0 , \\
m_{68}&=&m_{86}=-\frac{22\sqrt{2}}{243}\phi_0k^{3 - 2{\alpha^{\prime}}k^2 } , \\
m_{69}&=&m_{96}=-\frac{55}{729}\phi_0k^{3 - 2{\alpha^{\prime}}k^2 } , \\
m_{6,11}&=&m_{11,6}=\frac{11}{729}\phi_0k^{3 - 2{\alpha^{\prime}}k^2}\left(\frac{8}{5}{\alpha^{\prime}}k^2 - 25\right ) , \\
m_{6,12}&=&m_{12,6}=\frac{88}{729}\phi_0k^{3 - 2{\alpha^{\prime}}k^2}{\alpha^{\prime}}k^2 , \\ m_{77}&=&{\alpha^{\prime}}k^2 + 1+\frac{256}{243}\phi_0k^{3 - 2{\alpha^{\prime}}k^2 } , \\
m_{78}&=&m_{79}=m_{7,11}=m_{7,12}=m_{87}=m_{97}=m_{11,7}=m_{12,7}=0 , \\
m_{7,10}&=&m_{10,7}=-\frac{256\sqrt{2}}{729}\phi_0k^{3 - 2{\alpha^{\prime}}k^2 } , \\ m_{88}&=&\frac{{\alpha^{\prime}}k^2 + 1}{{\alpha^{\prime}}k^2}+\left(\frac{8}{81}+\frac{256}{243{\alpha^{\prime}}k^2}\right)\phi_0k^{3 - 2{\alpha^{\prime}}k^2 } , \\
m_{89}&=&m_{98}=\frac{10\sqrt{2}}{243}\phi_0k^{3 - 2{\alpha^{\prime}}k^2 } , \\
m_{8,10}&=&m_{10,8}=0 , \\ m_{8,11}&=&m_{11,8}=\frac{256\sqrt{2}}{3645}\phi_0k^{3 - 2{\alpha^{\prime}}k^2}-\frac{2\sqrt{2}}{243}\phi_0 k^{3 - 2{\alpha^{\prime}}k^2}\left(\frac{8}{5}{\alpha^{\prime}}k^2 - 25\right ) , \\
m_{8,12}&=&m_{12,8}=\frac{256\sqrt{2}}{729}\phi_0k^{3 - 2{\alpha^{\prime}}k^2}-\frac{16\sqrt{2}}{243}\phi_0 k^{3 - 2{\alpha^{\prime}}k^2}{\alpha^{\prime}}k^2 , \\ m_{99}&=&{\alpha^{\prime}}k^2 + 1+\frac{537}{729}\phi_0k^{3 - 2{\alpha^{\prime}}k^2 } , \\
m_{9,10}&=&m_{10,9}=0 , \\ m_{9,11}&=&m_{11,9}=\frac{1}{729}(-8{\alpha^{\prime}}k^2 + 125)\phi_0k^{3 - 2{\alpha^{\prime}}k^2 } , \\
m_{9,12}&=&m_{12,9}=-\frac{40}{729}{\alpha^{\prime}}k^2 \phi_0k^{3 - 2{\alpha^{\prime}}k^2 } , \\
m_{10,10}&=&\frac{2({\alpha^{\prime}}k^2 + 1)}{{\alpha^{\prime}}k^2}+\frac{512}{729}\phi_0k^{3 - 2{\alpha^{\prime}}k^2}+\frac{1024}{729{\alpha^{\prime}}k^2 } \phi_0k^{3 - 2{\alpha^{\prime}}k^2 } , \\
m_{10,11}&=&m_{11,10}=m_{10,12}=m_{12,10}=0 , \\ m_{11,11}&=&{\alpha^{\prime}}k^2 + 1+\frac{1}{729}\phi_0k^{3 - 2{\alpha^{\prime}}k^2}\left(-\frac{8}{5}{\alpha^{\prime}}k^2 + 25\right)^2 \\ & & { } + \frac{512}{729}\phi_0k^{3 - 2{\alpha^{\prime}}k^2}+\frac{512}{25\times 729}\phi_0k^{3 - 2{\alpha^{\prime}}k^2}{\alpha^{\prime}}k^2 , \\ m_{11,12}&=&m_{12,11}=-\frac{8}{729}{\alpha^{\prime}}k^2\left(-\frac{8}{5}{\alpha^{\prime}}k^2 + 25\right)\phi_0k^{3 - 2{\alpha^{\prime}}k^2 } + \frac{512}{3645}\phi_0k^{3 - 2{\alpha^{\prime}}k^2}{\alpha^{\prime}}k^2 , \\ m_{12,12}&=&{\alpha^{\prime}}k^2 + 1+\frac{1}{729}\phi_0k^{3 - 2{\alpha^{\prime}}k^2}(64 { \alpha^{\prime}}{}^2 ( k^2)^2 + 512 + 512{\alpha^{\prime}}k^2 ) , \\
m_{16}&=&m_{17}=m_{18}=m_{19}=m_{1,10}=m_{1,11}=m_{1,12}=m_{61 } \\ & = & m_{71}=m_{81}=m_{91}=m_{10,1}=m_{11,1}=m_{12,1}=0 , \\
m_{56}&=&m_{57}=m_{58}=m_{59}=m_{5,10}=m_{5,11}=m_{5,12}=m_{65 } \\ & = & m_{75}=m_{85}=m_{95}=m_{10,5}=m_{11,5}=m_{12,5}=0 , \\
m_{26}&=&m_{62}=m_{64}=m_{46}=\frac{11}{27}\phi_{\pm 1}k^{3 - 2{\alpha^{\prime}}k^2-\frac{2{\alpha^{\prime}}}{r^2 } } , \\
m_{27}&=&m_{72}=m_{47}=m_{74}=0 , \\
m_{28}&=&m_{82}=m_{84}=m_{48}=-\frac{2\sqrt{2}}{9}\phi_{\pm 1}k^{3 - 2{\alpha^{\prime}}k^2-\frac{2{\alpha^{\prime}}}{r^2 } } , \\
m_{29}&=&m_{92}=m_{94}=m_{49}=\frac{1}{27}\left(\frac{32{\alpha^{\prime}}}{r^2}-5\right)\phi_{\pm 1 } k^{3 - 2{\alpha^{\prime}}k^2-\frac{2{\alpha^{\prime}}}{r^2 } } , \\
m_{2,10}&=&-m_{10,2}=-m_{4,10}=m_{10,4}= -\frac{32 \sqrt{{\alpha^{\prime } } } i}{27 r}\phi_{\pm 1}k^{3 - 2{\alpha^{\prime}}k^2-\frac{2{\alpha^{\prime}}}{r^2 } } , \\
m_{2,11}&=&m_{11,2}=m_{11,4}=m_{4,11}=\frac{1}{27}\left(\frac{8}{5}{\alpha^{\prime}}k^2 - 25\right)\phi_{\pm 1 } k^{3 - 2{\alpha^{\prime}}k^2-\frac{2{\alpha^{\prime}}}{r^2 } } , \\
m_{2,12}&=&m_{12,2}=m_{12,4}=m_{4,12}=\frac{8}{27}{\alpha^{\prime}}k^2\phi_{\pm 1}k^{3 - 2{\alpha^{\prime}}k^2-\frac{2{\alpha^{\prime}}}{r^2 } } , \\
m_{36}&=&m_{63}=\frac{11}{27}\phi_0k^{3 - 2{\alpha^{\prime}}k^2}+\frac{19}{243}uk^{3 - 2{\alpha^{\prime}}k^2}+\frac{11}{729}\left\ { \left(16{\alpha^{\prime}}k^2-\frac{125}{2}\right)w-\frac{5}{2}v\right\ } k^{3 - 2{\alpha^{\prime}}k^2 } , \\
m_{37}&=&m_{73}=0 , \\
m_{38}&=&m_{83}=-\frac{2\sqrt{2}}{9}\phi_0k^{3 - 2{\alpha^{\prime}}k^2}-\frac{22\sqrt{2}}{243}uk^{3 - 2{\alpha^{\prime}}k^2 } + \frac{256\sqrt{2}}{729}w k^{3 - 2{\alpha^{\prime}}k^2 } \\ & & { } -\frac{2\sqrt{2}}{243}k^{3 - 2{\alpha^{\prime}}k^2 } \left\ { \left ( 16{\alpha^{\prime}}k^2-\frac{125}{2}\right ) w-\frac{5}{2}v\right\ } , \\
m_{39}&=&m_{93}=-\frac{5}{27}\phi_0k^{3 - 2{\alpha^{\prime}}k^2}-\frac{5}{729}k^{3 - 2{\alpha^{\prime}}k^2}\left\ { \left ( 16 { \alpha^{\prime}}k^2-\frac{125}{2}\right ) w-\frac{5}{2 } v\right\ } \\ & & { } + \frac{256}{729}vk^{3 - 2{\alpha^{\prime}}k^2 } -\frac{55}{729}u k^{3 - 2{\alpha^{\prime}}k^2 } , \\
m_{3,10}&=&m_{10,3}=0 , \\
m_{3,11}&=&m_{11,3}=\frac{1}{27}\left(\frac{8}{5}{\alpha^{\prime}}k^2 - 25\right)\phi_0k^{3 - 2{\alpha^{\prime}}k^2 } \\ & & { } + \frac{1}{729}\left(\frac{8}{5}{\alpha^{\prime}}k^2 - 25\right ) k^{3 - 2{\alpha^{\prime}}k^2}\left\{\left ( 16{\alpha^{\prime}}k^2 -\frac{125}{2}\right ) w-\frac{5}{2}v\right\ } \\ & & { } + \frac{256\times 5}{729}w k^{3 - 2{\alpha^{\prime}}k^2 } -\frac{512}{5\times 729}wk^{3 - 2{\alpha^{\prime}}k^2}{\alpha^{\prime}}k^2+\frac{11}{729}u k^{3 - 2{\alpha^{\prime}}k^2}\left(\frac{8}{5}{\alpha^{\prime}}k^2 -25\right ) , \\
m_{3,12}&=&m_{12,3}=\frac{8}{27}{\alpha^{\prime}}k^2\phi_0k^{3 - 2{\alpha^{\prime}}k^2}+\frac{8}{729}{\alpha^{\prime}}k^2 k^{3 - 2{\alpha^{\prime}}k^2 } \left\{\left ( 16{\alpha^{\prime}}k^2-\frac{125}{2}\right ) w-\frac{5}{2}v\right\ } \\ & & { } -\frac{512}{729}wk^{3 - 2{\alpha^{\prime}}k^2}{\alpha^{\prime}}k^2+\frac{88}{729}u k^{3 - 2{\alpha^{\prime}}k^2}{\alpha^{\prime}}k^2.\end{aligned}\ ] ] r. de mello koch , a. jevicki , m. mihailescu and r. tatar , lumps and @xmath34-branes in open string field theory , " _ phys.lett . _ * b482*(2000)249 - 254 , hep - th/0003031 ; + r. de mello koch and j.p .
rodrigues , lumps in level truncated open string field theory , " _ phys.lett .
_ * b495*(2000)237 - 244 , hep - th/0008053 ; + n. moeller , codimension two lump solutions in string field theory and tachyonic theories , " hep - th/0008101 .
n. berkovits , the tachyon potential in open neveu - schwarz string field theory , " _ jhep _ * 0004*(2000)022 , hep - th/0001084 ; + n. berkovits , a. sen and b. zwiebach , tachyon condensation in superstring field theory , " _ nucl
* b587*(2000)147 - 178 , hep - th/0002211 ; + p .- j .
de smet and j. raeymaekers , level four approximation to the tachyon potential in superstring field theory , " _ jhep _ * 0005*(2000)051 , hep - th/0003220 ; + a. iqbal and a. naqvi , tachyon condensation on a non - bps d - brane , " hep - th/0004015 ; + i. ya .
arefeva , a. s. koshelev , d. m. belov and p. b. medvedev , tachyon condensation in cubic superstring field theory , " hep - th/0011117 .
h. hata and s. shinohara , brst invariance of the non - perturbative vacuum in bosonic open string field theory , " _ jhep _ * 0009*(2000)035 , hep - th/0009105 ; + p. mukhopadhyay and a. sen , test of siegel gauge for the lump solution , " _ jhep _ * 0102*(2001)017 , hep - th/0101014 .
l. rastelli , a. sen and b. zwiebach , half - strings , projectors , and multiple d - branes in vacuum string field theory , " hep - th/0105058 ; + l. rastelli , a. sen and b. zwiebach , boundary cft construction of d - branes in vacuum string field theory , " hep - th/0105168 ; + l. rastelli , a. sen and b. zwiebach , vacuum string field theory , " hep - th/0106010 ; + v. a. kosteleck and r. potting , analytical construction of a nonperturbative vacuum for the open bosonic string , " _ phys . rev .
_ * d63*(2001)046007 , hep - th/0008252 ; + d.j .
gross and w. taylor ,
split string field theory i , " hep - th/0105059 ; + d.j .
gross and w. taylor ,
split string field theory ii , " hep - th/0106036 ; + t. kawano and k. okuyama ,
open string fields as matrices , " hep - th/0105129 ; + y. matsuo , bcft and sliver state , " hep - th/0105175 ; + y. matsuo , identity projector and d - brane in string field theory , " hep - th/0106027 ; + j.r .
excitations on wedge states and on the sliver , " hep - th/0105184 .
d. gross and a. jevicki , operator formulation of interacting string field theory(i ) , " _ nucl .
_ * b283*(1987)1 ; + e. cremmer , a. schwimmer and c. thorn , the vertex function in witten s formulation of string field theory , " _ phys . lett . _ * b179*(1986)57 ; + s. samuel , the physical and ghost vertices in witten s string field theory , " _ phys . lett . _ * b181*(1986)255 ; + n. ohta , covariant interacting string field theory in the fock - space representation , " _ phys . rev . _ * d34*(1986)3785 ; * d35*(1987)2627 ( erratum ) .
h. hata and s. teraguchi , test of the absence of kinetic terms around the tachyon vacuum in cubic string field theory , " hep - th/0101162 ; + i. ellwood and w. taylor ,
open string field theory without open strings , " hep - th/0103085 ; + b. feng , y .- h .
he and n. moeller , testing the uniqueness of the open bosonic string field theory vacuum , " hep - th/0103103 ; + i. ellwood , b. feng , y .- h .
he and n. moeller , the identity string field and the tachyon vacuum , " hep - th/0105024 .
a. recknagel and v. schomerus , boundary deformation theory and moduli spaces of d - branes , " _ nucl .
phys . _ * b545*(1999)233 , hep - th/9811237 ; + c. g. callan , i. r. klebanov , a. w. ludwig and j. m. maldacena , exact solution of a boundary conformal field theory , " _ nucl . phys . _ * b422*(1994)417 , hep - th/9402113 ; + j. polchinski and l. thorlacius , free fermion representation of a boundary conformal field theory , " _ phys .
rev . _ * d50*(1994)622 , hep - th/9404008 ; + a. sen ,
descent relations among bosonic d - branes , " _ int . j. mod
* a14*(1999)4061 - 4078 , hep - th/9902105 .
d. ghoshal and a. sen ,
tachyon condensation and brane descent relations in @xmath34-adic string theory , " _ nucl .
phys . _ * b584*(2000)300 - 312 , hep - th/0003278 ; + j. a. minahan , mode interactions of the tachyon condensate in @xmath34-adic string theory , " _ jhep _ * 0103*(2001)028 , hep - th/0102071 ; + j.a .
minahan , quantum corrections in p - adic string theory , " hep - th/0105312 .
minahan and b. zwiebach , field theory models for tachyon and gauge field string dynamics , " _ jhep _ * 0009*(2000)029 , hep - th/0008231 ; + j.a .
minahan and b. zwiebach , effective tachyon dynamics in superstring theory , " _ jhep _ * 0103*(2001)038 , hep - th/0009246 ; + j. a. minahan and b. zwiebach , gauge fields and fermions in tachyon effective field theories , " _ jhep _ * 0102*(2001)034 , hep - th/0011226 . | we study some properties of the tachyonic lumps in the level truncation scheme of bosonic cubic string field theory .
we find that several gauges work well and that the size of the lump as well as its tension is approximately independent of these gauge choices at level ( 2,4 ) .
we also examine the fluctuation spectrum around the lump solution , and find that a tachyon with @xmath0 and some massive scalars appear on the lump world - volume .
this result strongly supports the conjecture that a codimension 1 lump solution is identified with a d - brane of one lower dimension within the framework of bosonic cubic string field theory .
ut-944 + hep - th/0106068 + june , 2001 1.5 cm * * 1.5 cm + + |
in the past decade , energy self - localization in nonlinear dynamical lattices , leading to the formation of soliton - like _ intrinsic localized modes _ ( ilms ) , has become a topic of intense theoretical and experimental research .
much of this work has already been summarized in several reviews @xcite .
it was proposed that this mechanism would be relevant to a number of effects such as nonexponential energy relaxation in solids @xcite , local denaturation of the dna double strand @xcite , behavior of amorphous materials @xcite , propagation of light beams in coupled optical waveguides @xcite or the self - trapping of vibrational energy in proteins @xcite , among others .
ilms also have potential significance in some crystals , like acetanilide and related organics @xcite .
the theoretical efforts were complemented by a number of important experimental works suggesting the presence and importance of the ilms in magnetic @xcite and complex electronic materials @xcite , dna denaturation @xcite , as well as in coupled optical waveguide arrays @xcite and josephson ladders @xcite .
a ubiquitous model system for the study of ilms is the discrete nonlinear schrdinger ( dnls ) equation ( see e.g. , the review @xcite and references therein ) . within the framework of this model and , more generally , for klein - gordon lattices , it has recently been recognized that physically realistic setups require consideration of the ilm dynamics in higher spatial dimensions @xcite . in the most straightforward two - dimensional ( 2d ) case , almost all of these studies , with the exception of refs .
@xcite were performed for square lattices .
however , it was stressed in ref .
@xcite that non - square geometries may be relevant to a variety of applications , ranging from the explanation of dark lines in natural crystals of muscovite mica , to sputtering ( ejection of atoms from a crystal surface bombarded by high - energy particles ) , and , potentially , even to high - temperature superconductivity in layered cuprates .
besides that , it has been well recognized that triangular ( ta ) and hexagonal ( or _ honeycomb _ , hc ) lattices are relevant substrate structures in a number of chemical systems @xcite and , especially , in photonic band - gap ( pbg ) crystals @xcite .
notice that , in the context of the pbg crystals , the relevance of nonlinear effects has been recently highlighted for a square diatomic lattice @xcite .
the above discussion suggests the relevance of a systematic study of ilms in the paradigm dnls model for ta and hc lattices .
the aim of the present work is to address this issue ( including the stability of the ilm solutions ) , for the 2d lattices with both short - range and long - range interactions . in section
ii we discuss the effects of the non - square lattice geometry on the fundamental ilm state ( the one centered on a lattice site ) , and then explore effects of long - range interactions on this state . in section iii , we expand our considerations to other classes of solutions , which are either more general ones , such as twisted modes , which are also known in square lattices , or represent states that are specific to the ta and hc structures , viz . , discrete vortices .
we identify stable fundamental vortices in the ta and hc lattices with vorticity ( spin ) @xmath0 and @xmath1 , and with the hexagonal and honeycomb shape , respectively .
additionally , a triangular vortex is found in the hc lattice , but it is always unstable .
in this work we consider the two - dimensional dnls equation with the on - site cubic nonlinearity , @xmath2 the subscripts ( @xmath3 ) attached to the complex ( envelope ) field @xmath4 are two discrete spatial coordinates , @xmath5 is the constant of the linear coupling between nearest - neighbor sites , the summation over which is denoted by @xmath6 , and @xmath7 is the coordination number ( i.e. , the number of the nearest neighbors ) , which takes the values @xmath8 for the ta lattice ( see the left panel of fig . [ hfig1 ] ) , @xmath9 for the square lattice , and @xmath10 for the hc one ( see the right panel of fig . [ hfig1 ] ) .
the function @xmath11 represents a kernel of the long - range linear coupling , and @xmath12 is the lattice spacing .
it is worth noting here that the ta network is a simple bravais lattice with the coordinates of the grid nodes @xmath13 and @xmath14 ( see also ref .
the same is true for the most commonly used square lattice , which has @xmath15 , @xmath16 , but _ not _ for the hc structure , a simple representation of which ( for @xmath17 ) is @xmath18 , @xmath19 $ ] . more information on the latter structure ( which also represents , for instance , the arrangement of carbon atoms in a layer of graphite ) and its symmetries can be found in ref .
@xcite .
first , we will look for ilm solutions of the nearest - neighbor version of the model , setting @xmath20 .
stationary solutions with a frequency @xmath21 are sought for in the ordinary form ( see e.g. , ref .
@xcite ) , @xmath22 the substitution of eq .
( [ heq2 ] ) into eq .
( [ heq1 ] ) leads to a time - independent equation for the amplitudes @xmath23 .
the stationary solution being known , one can perform the linear - stability analysis around it in the same way as it has been done for the square lattice @xcite , assuming a perturbed solution in the form @xmath24 where @xmath25 is a perturbation with an infinitesimal amplitude @xmath26 .
deriving the leading - order equation for @xmath25 , and looking for a relevant solution to it in the form @xmath27 ( where the eigenfrequency @xmath28 is , generally speaking , complex ) , one arrives at an eigenvalue problem for @xmath29 : @xmath30 the inclusion of long - range effects into the linear - stability equations is straightforward . as the long - range coupling is accounted for by a linear operator acting on the complex field , terms @xmath31 and @xmath32 are to be added to eqs .
( [ heq4 ] ) and ( [ heq5 ] ) , respectively .
fundamental ( single - site - centered ) ilm solutions to the stationary equations were constructed by means of a newton - type method , adjusted to the non - square geometry of the ta and hc lattice . for the results presented
herein , we fix the frequency to be @xmath33 and vary the coupling constant @xmath5 , as one of the two parameters ( @xmath34 and @xmath5 ) can always be scaled out from the stationary equations .
we started from obvious single - site solutions ( with @xmath35 ) at the anti - continuum limit corresponding to @xmath36 @xcite , and then continued the solution to finite @xmath5 .
subsequently , the stability analysis was performed using eqs .
( [ heq4])-([heq5 ] ) for the corresponding lattice .
typical examples of stable and unstable fundamental ilms found in both the ta and hc lattices are displayed , by means of contour plots , in fig .
[ hfig4 ] .
the top panel of the figure shows , respectively , stable and unstable solutions , together with the associated spectral - plane diagrams ( showing the imaginary vs. real parts of the eigenfrequencies ) , for the ta lattice with @xmath37 ( top subplots ) and @xmath38 ( bottom subplots ) .
stable and unstable solutions in the hc lattice are shown in the bottom panel of fig .
[ hfig4 ] for @xmath37 ( top subplots ) and at @xmath39 ( bottom subplots ) .
a natural way to understand the stability of the ilms is to trace the evolution and bifurcations of the eigenfrequencies and associated eigenmodes with the increase of the nearest - neighbor coupling @xmath5 .
for the square lattice , we find , in line with results of refs .
@xcite , that a bifurcation generating an internal mode from the edge of the continuous spectrum ( the edge is at @xmath40 ) in the corresponding ilm occurs at a critical value @xmath41 . as the coupling is further increased , the pair of the corresponding eigenfrequencies moves towards the origin of the spectral plane , where they collide and bifurcate into an unstable pair of imaginary eigenfrequencies at @xmath42 ( i.e. , at @xmath43 in the present notation ) , so that the ilm in the square lattice is unstable for @xmath44 . in the ta and hc lattices ,
the scenario is found to be quite similar .
for the former lattice , the bifurcation of the two eigenfrequencies from the continuous band edge ( which is depicted by the dash - dotted line ) and their trajectory , as they change from real , i.e. , stable ( the solid line ) into imaginary , i.e. , unstable ( the dashed line ) , are shown in the top panel of fig .
[ hfig2 ] .
the bottom panel shows the same trajectory for the hc lattice . the pair of eigenvalues bifurcates at @xmath45 in the ta lattice , and they reach the origin , giving rise to the instability , at a point close to @xmath46 . in the hc lattice ,
the bifurcation giving rise to the originally stable eigenvalues occurs at @xmath47 ; they collide at the origin and become unstable at @xmath48 .
one can clearly identify the effect of geometry in these results .
in particular , since the instability occurs beyond the critical values of the coupling , it is the linear interaction between the neighbors that drives it .
consequently , since the coordination numbers for the different lattices are ordered as @xmath49 , the instability thresholds ( critical values of the coupling constant ) for these lattices should be ordered conversely , @xmath50 .
a similar understanding of the effect of the coordination number on the norm of the solution , @xmath51 , justifies the results displayed in fig .
[ hfig3 ] : the larger number of neighbors endows the ta branch ( solid line ) with a larger norm than the square one ( dash - dotted ) , which , in turn , has a larger norm than the hc lattice ( dashed ) . recently , a lattice model with a long - range coupling , which is relevant to magnon - phonon , magnon - libron and exciton - photon interactions , was introduced in ref . @xcite . in the framework of this model , it has been concluded that the relevant coupling kernel [ see eq.([heq1 ] ) ] is @xmath52 where @xmath53 is an amplitude of the kernel , @xmath54 measures the range of the interaction , and @xmath55 is the modified bessel function .
we will use this kernel below . inserting the kernel ( [ heq6 ] ) into eq .
( [ heq1 ] ) , one can see how the behavior of the branch is modified as a function of @xmath53 for a given fixed value of the nearest - neighbor coupling @xmath5 .
notice that similar results ( but on a logarithmic scale ) will be obtained if @xmath56 is varied , while @xmath57 is kept fixed , as it was detailed in ref .
thus , we fix @xmath58 hereafter . in fig .
[ hfig5 ] , we show the evolution of the numerically found internal - mode eigenvalues of ilm as a function of @xmath53 , for fixed @xmath37 . it can be observed that the increase of @xmath53 leads to an instability for @xmath59 .
the bottom panel shows the configuration and its internal - eigenmode frequency for @xmath60 , when the configuration is already unstable . carefully zooming into the ilm s in the case of long - range interactions ( data not shown here )
, one can notice a `` tail '' of the ilm , much longer than the size of the ilm in the case of the nearest - neighbor interaction , which is a natural consequence of the nonlocal character of the interaction in the present case .
for @xmath61 , we thus conclude that the long - range interaction `` cooperates '' with the short - range one , lowering the instability threshold . on the contrary ,
numerical results for @xmath62 show that the onset of the instability is _ delayed _ when the long- and short - range interactions compete with each other ( see also ref .
@xcite ) . one can extend the above considerations to the case where both @xmath53 and @xmath63 are varied and construct two - parameter diagrams , separating stability and instability regions .
an example is shown for the ta lattice in fig .
[ hfig5a ] . for a fixed @xmath5 ,
the critical value @xmath64 was identified , beyond which the ilm is unstable .
thus , in fig . [ hfig5a ] ilm s are stable below the curve and unstable above it .
we now turn our attention to solutions comprising many sites of the lattice . in this case too , the solutions are initially constructed in the anti - continuum limit @xmath36 , and then extended through continuation to finite values of @xmath5 .
firstly , we examine the so - called twisted localized modes ( tlms ) , which were originally introduced , in the context of 1d lattices , in refs .
later , they were studied in more detail in ref .
@xcite , and their stability was analyzed in ref .
they were subsequently used to construct topologically charged 2d solitons ( vortices ) in the square - lattice dnls equation in @xcite . in the case of the square lattice , and
subject to the same normalization as adopted above , i.e. , with @xmath33 , tlms are found to be stable for @xmath65 .
if the coupling exceeds this critical value , an oscillatory instability , which is manifested through a quartet of complex eigenvalues @xcite , arises due to the collision of the tlm s internal mode with the continuous spectrum ( the two have opposite _ krein signatures _
@xcite ) , as it has been detailed in ref .
the same scenario is found to occur in the ta lattice .
however , in the latter case the oscillatory instability sets in at @xmath66 , and the destabilization is a result of the collision of the eigenvalues with those that have ( slightly ) bifurcated from the continuous spectrum ( rather than with the edge of the continuous spectrum at @xmath67 , as in the square lattice ) .
similarly , in the hc lattice , the instability of tlms sets in at @xmath68 . notice that the instability thresholds follow the same ordering as the ones discussed in the previous section .
this can be justified by a similar line of arguments as given before .
the tlm in the ta lattice ( and its stability ) is displayed in the top panel of fig .
[ hfig6 ] for @xmath69 , which exceeds the instability threshold .
the bottom panel of the figure shows the variation ( as a function of the coupling @xmath5 ) of the critical eigenfrequency .
the real stable eigenfrequency , and the imaginary part of the unstable ones , after the threshold has been crossed , are shown , respectively , by solid and dashed lines .
figure [ hfig6a ] displays analogous results for the hc lattice .
the solution is shown at @xmath69 in the top panel .
it should be remarked that , in the 2d lattice , the tlms are solutions carrying vorticity ( topological charge ) @xmath70 @xcite ( although they are different from vortices proper , see below ) . the simplest way to see
this is by recognizing that tlm configurations emulate the continuum - limit expression @xmath71 , where @xmath72 is the angular coordinate in the 2d plane , i.e. , the real part of @xmath73 , the latter expression carrying vorticity @xmath74 .
it should also be added that , after the onset of the oscillatory instability , tlm solutions have been found to transform themselves into the fundamental ( single - site - centered ) ilm configurations , which is possible as the topological charge is not a dynamical invariant in lattices @xcite .
going beyond tlms , it is appropriate to consider possible lattice solitons which conform to the symmetry of the ta or hc lattice .
in fact , these are the most specific dynamical modes supported by the lattices .
an example of this sort in the ta lattice are `` hexagonal '' ilms shown in fig .
[ hfig8 ] .
the top panel of the figure shows the profile of these modes in the anti - continuum limit , and the bottom panel displays two actual examples of these modes .
the top subplot shows the hexagonal ilm for @xmath75 , when it is stable , while the bottom subplot shows the mode at @xmath76 , after the onset of three distinct oscillatory instabilities .
the first and second instabilities set in at @xmath77 and @xmath78 respectively , while the final eigenvalue quartet appears at @xmath79 . measuring the topological charge of this solution around the contour in the top panel of fig .
[ hfig7 ] , we find ( since each jump from @xmath80 to @xmath81 can be identified as a @xmath82 phase change ) that the whole solution has a total phase change of @xmath83 , hence its topological charge ( vorticity , or `` spin '' ) is @xmath0 .
then , the presence of three oscillatory instabilities agrees with a recent conjecture @xcite , which states that the number of negative - krein - sign eigenvalues ( and hence the number of potential oscillatory instabilities ) coincides with the topological charge of the 2d lattice soliton .
however , if one examines more carefully the stability picture , one finds that , due to the symmetry of the solution , two of these eigenvalues have multiplicity @xmath84 .
hence , the conjecture needs to be refined , to take into regard the potential presence of symmetries .
the thus revised conjecture states that the topological charge of the solution should be equal to the geometric ( but not necessarily algebraic ) multiplicity of the eigenvalues with negative krein signature .
it is natural to ask then to what configuration this hexagonal ilm will relax once it becomes unstable . to address the issue , we performed direct numerical simulations for @xmath76 .
results are shown in our subplots of fig .
[ hfig8 ] .
in particular , the top left panel shows the solution at @xmath85 ( the configuration at @xmath86 was the unstable hexagonal ilm ) .
it can clearly be observed that the instability that sets in around @xmath87 ( according to the other three subplots ) transforms the hexagonal vortex into a fundamental ( zero - vorticity ) ilm ; recall that such an outcome of the instability development is possible because the vorticity is _ not _ conserved in lattice systems @xcite . for the hc lattice ,
a vortex soliton of a triangular form was found , see an example in fig .
[ hfig9 ] for @xmath88 .
we have found that this solution is unstable for _ all values _ of @xmath5 .
another vortex soliton , with a honeycomb shape , was also found in the hc lattice , see fig .
[ hfig10 ] .
this one is _
stable _ at a sufficiently weak coupling .
if a contour is drawn around this solution ( shown in the top panel of the figure for the anti - continuum limit ) , the net phase change is found to be @xmath89 , hence the corresponding topological charge is @xmath1 . in accordance with
the conjecture mentioned above , when the solution is stable , we find five internal modes with negative krein signature .
these modes eventually lead , as the coupling is increased , to five oscillatory instabilities . in the bottom panel of fig .
[ hfig10 ] , the top subplot shows the case with @xmath90 , when the honeycomb - shaped vortex soliton in the hc lattice is linearly stable .
the bottom subplot features the presence of five eigenvalue quartets in the case @xmath91 , when all five oscillatory instabilities have been activated .
the first instability occurs at @xmath92 , the second at @xmath93 , the third at @xmath94 , the fourth at @xmath95 , and the fifth sets in at @xmath96 . to illustrate the result of the development of the instability of the latter vortex in the case in which it is unstable
, we have again resorted to direct numerical integration of eq .
( [ heq1 ] ) .
an example is shown in fig .
[ hfig11 ] for @xmath91 . at @xmath97 ,
only one main pulse is sustained .
hence , in this case too , the multiply charged topological soliton is transformed , through the instability , into the stable ( for this value of @xmath5 ) fundamental zero - vorticity ilm .
in this work , we have studied a paradigm nonlinear lattice dynamical model , namely the dnls equation , in two spatial dimensions for non - square lattices .
the triangular and honeycomb networks were considered , as the most important examples of bravais and non - bravais 2d lattices , which are relevant to chemical and optical applications . in the case of nearest - neighbor interactions , it was found that the instability thresholds for the fundamental solution centered at a single lattice site depend on the coordination number .
the instability appears in the triangle lattice at a smaller value of the coupling constant than in the square lattice , while the opposite is true for the honeycomb lattice .
the effect of the long - range interactions was also examined in this context .
it was found that these interactions accelerate or delay the onset of the instability if they have the same sign as the nearest - neighbor coupling , or the opposite sign .
diagrams in the two - parameter plane were constructed , identifying regions of stability and instability in the presence of both the short- and long - range coupling .
more complicated lattice solitons , which essentially extend to several lattice sites , were also examined .
a prototypical example of the extended solitons are twisted modes , for which the phenomenology was found to be similar to that in the square lattice , but with , once again , appropriately shifted thresholds .
we have also examined solutions with a higher topological charge , which play the role of fundamental vortices in the triangular and honeycomb lattices , their vorticity being , respectively , @xmath0 and @xmath1 .
stability of these vortices was studied in detail .
when instabilities occurred , their outcome was examined by means of direct time integration , showing the transformation into a simple fundamental soliton with zero vorticity .
further steps in the study of localized modes in these nonlinear lattices may address traveling discrete solitons , as well as generalization to the three - dimensional case . in terms of applications , a relevant object are nonlinear photonic band - gap crystals based on non - square lattices .
van der meer , nonlinearity * 3 * , 1041 ( 1990 ) ; i.v .
barashenkov , d.e .
pelinovsky and e.v .
zemlyanaya , phys .
* 80 * , 5117 ( 1998 ) ; a. de rossi , c. conti and s. trillo , phys .
lett . * 81 * , 85 ( 1998 ) . | we study the existence and stability of localized states in the discrete nonlinear schrdinger equation ( dnls ) on two - dimensional non - square lattices .
the model includes both the nearest - neighbor and long - range interactions . for the fundamental strongly localized soliton ,
the results depend on the coordination number , i.e. , on the particular type of the lattice .
the long - range interactions additionally destabilize the discrete soliton , or make it more stable , if the sign of the interaction is , respectively , the same as or opposite to the sign of the short - range interaction .
we also explore more complicated solutions , such as twisted localized modes ( tlm s ) and solutions carrying multiple topological charge ( vortices ) that are specific to the triangular and honeycomb lattices . in the cases when such vortices are unstable
, direct simulations demonstrate that they turn into zero - vorticity fundamental solitons . |
turbulent convection in a fluid heated from below , known as rayleigh - bnard convection ( rbc ) , has been under intense study for some time [ for reviews , see _
e.g. _ @xcite ] .
a central prediction of models for this system [ @xcite ] is the heat transported by the fluid .
it is usually described in terms of the nusselt number @xmath25 where @xmath26 is the heat current , @xmath3 the cell height , @xmath27 the cross - sectional area , @xmath28 the thermal conductivity , and @xmath29 the applied temperature difference .
the nusselt number depends on the rayleigh number @xmath30 and on the prandtl number @xmath31 here @xmath32 is the isobaric thermal expansion coefficient , @xmath33 the acceleration of gravity , @xmath34 the thermal diffusivity , and @xmath35 the kinematic viscosity .
an important feature of turbulent rbc is the existence of a large - scale circulation ( lsc ) of the fluid [ @xcite ] . for cylindrical samples of aspect ratio @xmath36
the lsc is known to consist of a single cell , with fluid rising along the wall at some azimuthal location @xmath37 and descending along the wall at a location @xmath38 [ see , for instance , @xcite ] . as @xmath39 decreases ,
the nature of the lsc is believed to change .
for @xmath40 it is expected [ @xcite ] that the lsc consists of two or more convection cells , situated vertically one above the other .
regardless of the lsc structure , the heat transport in turbulent rbc is mediated by the emission of hot ( cold ) volumes of fluid known as plumes " from a more or less quiescent boundary layer above ( below ) the bottom ( top ) plate .
these plumes are swept away laterally by the lsc and rise ( fall ) primarily near the side wall .
their buoyancy helps to sustain the lsc . in a recent paper @xcite
reported measurements using a cylindrical sample of water with @xmath41 and with @xmath42 m and @xmath10 m for @xmath43 .
their sample thus had an aspect ratio @xmath9 at the borderline between a single - cell and a multi - cell lsc .
they found exceptionally long relaxation times of @xmath15 that they attributed to a switching of the lsc structure between two states .
multi - stability was observed also in nusselt - number measurements by @xcite for a @xmath44 sample ( see also @xcite for a discussion of these data ) .
chill et al . also found that @xmath15 was reduced by tilting the sample through an angle @xmath0 relative to gravity by an amount given approximately by @xmath45 when @xmath0 is measured in radian .
a reduction by two to five percent of @xmath15 ( depending on @xmath1 ) due to a tilt by @xmath46 of a @xmath44 sample was reported as well recently by @xcite , although in that paper the @xmath0-dependence of this effect was not reported .
chill et al .
developed a simple model that yielded a depression of @xmath15 for the two - cell structure that was consistent in size with their measurements .
their model also assumes that no depression of @xmath15 should be found for a sample of aspect ratio near unity where the lsc is believed to consist of a single convection cell ; they found some evidence to support this in the work of @xcite .
indeed , recent measurements by @xcite for @xmath47 gave the same @xmath15 within 0.1 percent for a level sample and a sample tilted by 0.035 rad .
in this paper we report on a long - term study of rbc in a cylindrical sample with @xmath48 .
as expected , we found no long relaxation times because the lsc is uniquely defined .
the establishment of a statistically stationary state after a large change of @xmath1 occurred remarkably quickly , within a couple of hours , and thereafter there were no further long - term drifts over periods of many days .
we also studied the orientation @xmath49 of the circulation plane of the lsc by measuring the side - wall temperature at eight azimuthal locations [ @xcite ] . with
the sample carefully leveled ( _ i.e. _ @xmath24 ) we found @xmath49 to change erratically , with large fluctuations . there were occasional relatively rapid reorientations , as observed before by @xcite .
the reorientations usually consisted of relatively rapid rotations , and rarely were reversals involving the cessation of the lsc followed by its re - establishment with a new orientation .
this lsc dynamics yielded a broad probability distribution - function @xmath50 , although a preferred orientation prevailed .
when the sample was tilted relative to gravity through an angle @xmath0 , a well defined new orientation of the lsc circulation plane was established , @xmath50 became much more narrow , and virtually all meandering and reorientation of the lsc was suppressed .
we found that @xmath15 was reduced very slightly by tilting the sample .
we obtained @xmath14 $ ] .
this effect is about a factor of 50 smaller than the one observed by chill et al . for their @xmath44 sample . from side - wall - temperature measurements at two opposite locations we determined time cross - correlation functions @xmath51 .
the @xmath51 had a peak that could be fitted well by a gaussian function , centered about a characteristic time @xmath52 that we interpreted as corresponding to the transit time needed by long - lived thermal disturbances to travel with the lsc from one side of the sample to the other , i.e. to half a turnover time of the lsc .
we found that the @xmath0-dependence of the corresponding reynolds number @xmath17 is given by @xmath18 $ ] .
a similar result was obtained from the auto - correlation functions of individual thermometers .
thus there is an @xmath53 effect of @xmath0 on @xmath21 , and yet the effect of @xmath0 on @xmath15 was seen to be nearly two orders of magnitude smaller .
we also determined the temperature amplitude @xmath19 of the azimuthal temperature variation at the mid - plane .
we expect @xmath19 to be a monotonically increasing function of the speed of the lsc passing the mid - plane , i.e. of the reynolds number .
we found @xmath20 $ ] .
thus , for small @xmath0 its @xmath0-dependence is very similar to that of the reynolds number . from the large effect of @xmath0 on @xmath21 and the very small effect on @xmath15
we come to the important conclusion that the heat transport in this system is not influenced significantly by the strength of the lsc .
this heat transport thus must be determined primarily by the efficiency of instability mechanisms in the boundary layers .
it seems reasonable that these mechanisms should be nearly independent of @xmath0 when @xmath0 is small .
this result is consistent with prior measurements by @xcite , who studied the lsc and the nusselt number in a sample with a rectangular cross section .
they inserted vertical grids above ( below ) the bottom ( top ) plate that suppressed the lsc , and found that within their resolution of a percent or so the heat transport was unaltered .
their shadowgraph visualizations beautifully illustrate that the plumes are swept along laterally by the lsc when there are no grids and rise or fall vertically due to their buoyancy in the presence of the grids .
@xcite also studied the effect of tilting their rectangular sample by an angle of 0.17 rad .
consistent with the very small effect of tilting on @xmath15 found by us , they found that within their resolution the heat transport remained unaltered .
we observed that the sudden reorientations of the lsc that are characteristic of the level sample are strongly suppressed by even a small tilt angle .
for the present work we used the large " and the medium " sample and apparatus described in detail by @xcite .
copper top and bottom plates each contained five thermistors close to the copper - fluid interface .
the bottom plate had imbedded in it a resistive heater capable of delivering up to 1.5 kw uniformly distributed over the plate .
the top plate was cooled via temperature - controlled water circulating in a double - spiral channel . for the nusselt - number measurements a temperature set - point for a digital feedback regulator
was specified .
the regulator read one of the bottom - plate thermometers at time intervals of a few seconds and provided appropriate power to the heater .
the top - plate temperature was determined by the temperature - controlled cooling water from two neslab rte740 refrigerated circulators .
each apparatus was mounted on a base plate that in turn was supported by three legs consisting of long threaded rods passing vertically through the plate .
the entire apparatus thus could be tilted by an angle @xmath0 relative to the gravitational acceleration by turning one of the rods . the maximum tilt angle attainable was 0.12 ( 0.21 ) rad for the large ( medium ) sample .
the nusselt number was calculated using the temperatures recorded in each plate and the power dissipated in the bottom - plate heater .
the side wall was plexiglas of thickness 0.64 cm ( 0.32 cm ) for the large ( medium ) sample .
it determined the length @xmath3 of the sample . around a circumference
the height was uniformly @xmath54 cm ( @xmath55 cm ) for the large ( medium ) sample .
the inside diameter was @xmath56 cm ( @xmath57 cm ) for the large ( medium ) sample .
the end plates had anvils that protruded into the side wall , thus guaranteeing a circular cross section near the ends .
for the large sample we made measurements of the outside diameter near the half - height after many months of measurements and found that this diameter varied around the circumference by less than 0.1% .
imbedded in the side wall and within 0.06 cm of the fluid - plexiglas interface were eight thermistors , equally spaced azimuthally and positioned vertically at half height of the sample .
they yielded a relatively high ( low ) temperature reading at the angular positions where there was up - flow ( down - flow ) of the lsc .
a fit of @xmath58 yielded the mean center temperature @xmath59 , the angular orientation @xmath49 of the lsc ( relative to the location of thermistor 0 ) , and a measure @xmath19 of the lsc strength .
we expect the size of @xmath19 to be determined by the heat transport across a viscous boundary layer separating the lsc from the side wall . thus @xmath19 should be a monotonically increasing function of the lsc reynolds number @xmath21 because the boundary - layer thickness is expected to decrease with @xmath21 as @xmath60 , and because the azimuthal temperature variation carried by the lsc near the boundary layer increases with @xmath1 and thus with @xmath21 .
however , the precise relationship between @xmath19 and @xmath21 is not obvious .
experimentally we find , over the range @xmath61 and for the large sample , that @xmath19 is related to @xmath1 by an effective power law @xmath62 , whereas @xmath63 in this range , yielding @xmath64 .
we would then expect that @xmath19 and @xmath21 will have a similar dependence on @xmath0 ( at least for small @xmath0 ) , albeit possibly with somewhat different coefficients .
from time series of the @xmath65 taken at intervals of a few seconds and covering at least one day we determined the cross - correlation functions @xmath66 corresponding to signals at azimuthal positions displaced around the circle by @xmath67 ( _ i.e. _ @xmath68 ) .
these functions are given by @xmath69\times[t_j(t+\tau ) -
\langle t_j(t)\rangle_t ] \rangle_t\ .
\label{eq : corfunc}\ ] ] we also calculated the auto - correlation functions corresponding to @xmath70 in eq .
[ eq : corfunc ] , for all eight thermometers .
initially each sample was carefully leveled so that the tilt angle relative to gravity was less than @xmath71 radian .
later it was tilted deliberately to study the influence of a non - zero @xmath0 on the heat transport .
the fluid was water at 40@xmath72c where @xmath73 k@xmath74 , @xmath75 @xmath76/s , and @xmath77 @xmath76/s , yielding @xmath8 .
in fig . [ fig : transients]a we show the initial evolution of the top and bottom temperatures of the large sample in a typical experiment .
initially the heat current was near zero and @xmath78 and @xmath79 were close to 40@xmath72c .
the sample had been equilibrated under these conditions for over one day . near @xmath80 h a new temperature set point of 50@xmath72c was specified for the bottom plate , and the circulator for the top plate was set to provide @xmath81c . from fig .
[ fig : transients]a one sees that there were transients that lasted until about 0.9 h ( 1.2 h ) for @xmath78 ( @xmath79 ) .
these transients are determined by the response time and power capability of the bottom - plate heater and the top - plate cooling water and are unrelated to hydrodynamic phenomena in the liquid .
figures [ fig : transients]b and c show the evolution of the heat current . after the initial rapid rise until @xmath82 h the current slowly evolved further to a statistically stationary value until @xmath83 h. a fit of the exponential function @xmath84 to the data for @xmath85 h is shown by the solid line in fig .
[ fig : transients]c and yielded a relaxation time @xmath86 h. we attribute this transient to the evolution of the fluid flow .
it is interesting to compare @xmath87 with intrinsic time scales of the system .
the vertical thermal diffusion time @xmath88 is 467 hours .
obviously it does not control the establishment of the stationary state .
if we consider that it may be reduced by a factor of @xmath89 with @xmath90 , we still obtain a time sale of 1.78 hours that is longer than @xmath87 .
we believe that the relatively rapid equilibration is associated with the establishment of the top and bottom boundary layers that involve much shorter lengths @xmath91 and @xmath92 .
it also is necessary for the large - scale circulation to establish itself ; but , as we shall see , its precise reynolds number is unimportant for the heat transport .
in addition , the lsc can be created relatively fast since this is not a diffusive process .
figure [ fig : transients ] shows the behavior of the system only during the first six hours and does not exclude the slow transients reported by chill _
et al . _ that occurred over time periods of @xmath93 hours .
thus we show in fig .
[ fig : n_of_t ] results for @xmath94 from a run using the large sample that was continued under constant externally imposed conditions for nine days .
each point corresponds to a value of @xmath15 based on a time average over two hours of the plate temperatures and the heat current .
note that the vertical range of the entire graph is only 0.8 % .
thus , within a small fraction of 1 % , the results are time independent .
indeed , during nearly a year of data acquisition for a @xmath47 sample at various rayleigh numbers , involving individual runs lasting from one to many days , we have never experienced long - term drifts or changes of @xmath15 after the first few hours .
this differs dramatically from the observations of chill _
et al_. who found changes by about 2 % over about 4 days .
we conclude that the slow transients observed by them for their @xmath44 sample do not occur for @xmath48 .
to document further the stationary nature of the system , we compared results from the large sample for @xmath15 obtained from many runs , each of one to ten days duration , over a period of about five months [ @xcite ] .
the scatter of the data at a given @xmath1 is only about 0.1% .
this excellent reproducibility would not be expected if there were slow transients due to transitions between different states of the lsc .
although work in our laboratory with other aspect ratios has been less extensive , we also have not seen any evidence of drifts or transients for the larger @xmath95 and 6 [ @xcite ] nor for the smaller @xmath96 and 0.28 [ @xcite ] .
it may be that @xmath44 , being near the borderline between a single - cell lsc and more complicated lsc structures [ @xcite ] , is unique in this respect . in fig .
[ fig : n_compare ] we compare results for @xmath97 from our large sample [ @xcite ] with those reported by chill _
et al . _ ( stars ) .
our results are larger by about 15% . to find a reason for this difference
, we first look at the @xmath39 and @xmath98 dependence .
the open ( solid ) circles represent our data for @xmath8 and @xmath99 and show that the dependence of @xmath15 on @xmath39 is not very strong .
the open squares ( diamonds ) are our results for @xmath100 and @xmath101 and indicate that @xmath15 actually _ increases _ slightly with @xmath98 .
thus the lower values of @xmath15 ( compared to ours ) obtained by chill et al .
for @xmath102 and @xmath44 can not be explained in terms of the @xmath39 and @xmath98 dependence of @xmath15 . some of the difference can be attributed to non - boussinesq effects that tend to reduce @xmath15 [ @xcite ] .
however , for the largest @xmath29 used by chill et al .
( 31@xmath72c ) we expect this effect to be somewhat less than 1 % [ @xcite ] .
finally , the effect of the finite conductivity of the top and bottom plates comes to mind .
this can reduce @xmath15 by several % when @xmath29 is large [ @xcite ] , but it is difficult to say precisely by how much .
it seems unlikely that this effect can explain the entire difference , particularly at the smaller @xmath1 ( and thus @xmath29 ) where it is relatively small .
in fig . [ fig : n_tiltresponse ] we show results for @xmath15 from the large sample at @xmath103 . each data point was obtained from a two - hour average of measurements of the various temperatures and of @xmath26 .
three data sets , taken in temporal succession , for tilt angles @xmath104 , and 0.122 are shown .
all data were normalized by the mean of the results for @xmath24 .
typically , the standard deviation from the mean of the data at a given @xmath0 was 0.13% .
the vertical dotted lines and the change in the data symbols show where @xmath0 was changed .
one sees that tilting the cell caused a small but measurable reduction on @xmath15 . in fig .
[ fig : n_of_beta ] we show the mean value for each tilt angle , obtained from runs of at least a day s duration a each @xmath0 , as a function of @xmath105 .
one sees that @xmath15 decreases linearly with @xmath0 .
a fit of a straight line to the data yielded @xmath106 \ .
\label{eq : n}\ ] ] with @xmath107 .
simlar results for the medium cell are compared with the large - cell results in fig .
[ fig : nred_of_beta ] . at the smaller rayleigh number of the medium sample the effect of @xmath0 on @xmath15 is somewhat less .
because the effect of @xmath0 on @xmath15 is so small , we did not make a more detailed investigation of its rayleigh - number dependence .
chill et al . proposed a model that predicts a significant tilt - angle effect on @xmath15 for @xmath44 where they assume the existence of two lsc cells , one above the other .
they also assumed that there would be no effect for @xmath47 where there is only one lsc cell .
although we found an effect for our @xmath47 sample , we note that it is a factor of about 50 smaller than the effect observed by chill et al . for @xmath44 .
in fig . [ fig : theta+delta ] we show the angular orientation @xmath49 ( a ) and the temperature amplitude @xmath19 ( b ) of the lsc . for the first 8000 sec shown in the figure , the sample was level ( @xmath108 ) .
one sees that @xmath49 varied irregularly with time .
the probability - distribution function @xmath50 is shown in fig .
[ fig : p_of_theta ] as solid dots .
essentially all angles are sampled by the flow , but there is a preferred direction close to @xmath109 . at t
= 8000 sec , the sample was tilted through an angle @xmath110 radian .
the direction of the tilt was chosen deliberately so as to oppose the previously prevailing preferred orientation . as a consequence one sees a sharp transition with a change of @xmath49 by
approximately @xmath67 .
the temperature amplitude @xmath19 on average increased slightly , and certainly remained non - zero . from this
we conclude that the transition took place via rotation of the lsc , and not by cessation that would have involved a reduction of @xmath19 to zero [ see @xcite ] .
we note that @xmath111 fluctuated much less after the tilt .
the results for @xmath50 after the tilt are shown in fig .
[ fig : p_of_theta ] as open circles .
they confirm that the maximum was shifted close to @xmath112 , and that the distribution was much more narrow . in fig .
[ fig : pshift_of_theta ] we show @xmath50 for @xmath113 ( solid circles ) , 0.044 ( open circles ) , and 0.026 ( solid squares ) .
one sees that a reduction of @xmath0 leads to a broadening of @xmath50 .
the square root of the variance of data like those in fig .
[ fig : pshift_of_theta ] is shown in fig .
[ fig : dp_of_theta ] on a logarithmic scale as a function of @xmath0 on a linear scale .
even a rather small tilt angle caused severe narrowing of @xmath50 . in fig .
[ fig : delta ] we show the temperature - amplitude @xmath114 of the lsc as a function of @xmath0 . as was the case for @xmath15 ,
the data are averages over the duration of a run at a given @xmath0 ( typically a day or two ) .
the solid ( open ) circles are for positive ( negative ) @xmath0 . the data can be represented well by either a linear or a quadratic equation .
a least - squares fit yielded @xmath115 \label{eq : delta}\ ] ] with @xmath116 k. using eq .
[ eq : corfunc ] , we calculated the auto - correlation functions ( ac ) @xmath117 , as well as the cross - correlation functions ( cc ) @xmath118 , of the temperatures measured on opposite sides of the sample .
typical examples are shown in fig .
[ fig : cor ] .
the cc has a characteristic peak that we associate with the passage of relatively hot or cold volumes of fluid at the thermometer locations .
such temperature cross - correlations have been shown , _
e.g. _ by @xcite and @xcite , to yield delay times equal to those of velocity - correlation measurements , indicating that warm or cold fluid volumes travel with the lsc .
the function @xmath119\ , \label{eq : ccfit}\ ] ] consisting of an exponentially decaying background ( that we associate with the random time evolution of @xmath49 ) and a gaussian peak , was fitted to the data for the cc .
the fitted function is shown in fig .
[ fig : cor ] as a solid line over the range of @xmath120 used in the fit .
it is an excellent representation of the data and yields the half turnover time @xmath121 of the lsc .
similarly , we fitted the function @xmath122 + b_2 exp \left [ -\left ( \frac{\tau - t^{i , i}_2}{\tau^{i , i}_2}\right ) ^2 \right ] \label{eq : acfit}\ ] ] to the ac data .
it consists of two gaussian peaks , one centered at @xmath123 and the other at @xmath124 , and the exponential background .
we interpret the location @xmath125 of the second gaussian peak to correspond to a complete turnover time @xmath126 of the lsc . in terms of the averages @xmath127 and
@xmath128 over all 8 thermometers or thermometer - pair combinations we define [ @xcite ] the reynolds numbers @xmath129 and @xmath130 here the length scale @xmath131 was used to convert the turnover time @xmath132 into a lsc speed @xmath133 . for @xmath47 , the length @xmath134 might have been used instead , as was done for instance by @xcite .
this would have led to a reynolds number larger by a factor of two . in fig .
[ fig : re]a and b we show @xmath135 and @xmath136 respectively .
the solid circles are for positive and the open ones for negative @xmath0 .
one sees that @xmath135 and @xmath136 initially grow linearly with @xmath0 , but the data also reveal some curvature as @xmath105 becomes larger .
thus we fitted quadratic equations to the data and obtained @xmath137 \label{eq : recc}\ ] ] and @xmath138 \label{eq : reac}\ ] ] with @xmath139 and @xmath140 ( all parameter errors are 67% confidence limits ) .
the results for @xmath141 and @xmath142 are about 10% higher than the prediction by @xcite for our @xmath98 and @xmath1 .
the excellent agreement between @xmath17 and @xmath143 is consistent with the idea that the cc yields @xmath144 and that the ac gives @xmath132 . as expected ( see sect . [
sec : apparatus ] ) , the @xmath0-dependences of both reynolds numbers are the same within their uncertainties .
it is interesting to see that the coefficients of the linear term also agree with the corresponding coefficient for @xmath19 ( eq . [ eq : delta ] ) .
this suggests that there may be a closer relationship between @xmath145 and @xmath21 than we would have expected _ a priori_. however , the coefficient of the linear term in eq .
[ eq : recc ] or [ eq : reac ] is larger by a factor of about 50 than the corresponding coefficient for the nusselt number in eq .
[ eq : n ] . in fig .
[ fig : re_delta_ratios ] we show measurements of @xmath17 and of @xmath19 , each normalized by its value at @xmath24 , as a function of @xmath0 for the medium sample and @xmath146 . for this sample
we were able to attain larger values of @xmath0 than for the large one .
one sees that @xmath19 and @xmath17 have about the same @xmath0 dependence for small @xmath0 , but that @xmath19 then increases more rapidly than @xmath17 as @xmath0 becomes large .
although we do not know the reason for this behavior , it suggests that the larger speed of the lsc enhances the thermal contact between the side wall and the fluid interior . the rayleigh - number dependence of @xmath17 at constant @xmath0 is shown in fig .
[ fig : re_ratios ] . here
the open ( solid ) circles are from the medium ( small ) sample
. there is consistency between the two samples , and the data can be described by a power law with a small negative exponent .
the solid line is drawn to correspond to an exponent of @xmath23 . as seen in fig .
[ fig : p_of_theta ] , the lsc assumes an orientation for which gravity enhances the velocity above ( below ) the bottom ( top plate ) , _ i.e. _ the lsc flows uphill " at the bottom where it is relatively warm and downhill " at the top where it is relatively cold .
this leads to an enhancement of the reynolds number of the lsc .
as suggested by @xcite , one can model this effect by considering the buoyancy force per unit area parallel to the plates .
this force can be estimated to be @xmath147 where @xmath148 is the boundary - layer thickness .
it is opposed by the increase of the viscous shear stress across the boundary layer that may be represented by @xmath149 where @xmath150 is the extra speed gained by the lsc due to the tilt .
equating the two , substituting @xmath151 solving for @xmath150 , using eq .
[ eq : r ] for @xmath1 , and defining @xmath152 one obtains @xmath153 for the enhancement of the reynolds number of the lsc . from our measurements at large @xmath1 we found that @xmath21 [ @xcite ] and @xmath15 [ @xcite ] can be represented within experimental uncertainty by @xmath154 giving @xmath155 for our @xmath8 and @xmath103 one finds @xmath156 , compared to the experimental value @xmath157 from @xmath17 [ eq . [ eq : recc ] ] and @xmath158 from @xmath143 [ eq . [ eq : reac ] ] .
we note that the coefficient @xmath159 in eq .
[ eq : re ] depends on the definition of @xmath21 given in eqs .
[ eq : re_cc ] an [ eq : re_ac ] that was used in deriving the result eq .
[ eq : re_exp ] .
if the length scale @xmath134 had been used instead of @xmath131 to define the speed of the lsc , as was done for instance by @xcite , this coefficient would have been smaller by a factor of two , yielding near - perfect agreement with the measurements . in fig .
[ fig : re_ratios ] one sees that the predicted dependence on @xmath160 also is in excellent agreement with the experimental results . however , such good agreement may be somewhat fortuitous , considering the approximations that were made in the model .
particularly the use of eq .
[ eq : bl ] for the boundary - layer thickness is called into question at a quantitative level by measurements of @xcite that revealed a significant variation of @xmath161 with lateral position . in addition
, it is not obvious that the thermal boundary - layer thickness @xmath161 should be used , as suggested by @xcite , to estimate the shear stress ; perhaps the thickness of the viscous bl would be more appropriate . in discussing their @xmath44 sample
, @xcite took the additional step of assuming that the relative change due to a finite @xmath0 of @xmath15 is equal to the relative change of @xmath21 . for our sample with @xmath47 this assumption does not hold . as we saw above ,
the relative change of @xmath15 is a factor of about 50 less than the relative change of @xmath21 .
the origin of the ( small ) depression of the nusselt number is not so obvious .
naively one might replace @xmath33 in the definition of the rayleigh number by @xmath162 ; but this would lead to a correction of order @xmath163 whereas the experiment shows that the correction is of order @xmath0 , albeit with a coefficient that is smaller than of order one .
the linear dependence suggests that the effect of @xmath0 on @xmath15 may be provoked by the change of @xmath21 with @xmath0 , but not in a direct causal relationship .
it is known from direct numerical simulation [ @xcite ] and from several experiments [ @xcite ] that the lsc can undergo relatively sudden reorientations .
not unexpectedly , we find that the tilt angle strongly influences the frequency of such events . for a level sample ( @xmath24 ) we demonstrated elsewhere [ @xcite ] that reorientations can involve changes of the orientation of the plane of circulation of the lsc through any angular increment @xmath164 , with the probability @xmath165 increasing with decreasing @xmath164 .
thus , in order to define a reorientation " , we established certain criteria .
we required that the magnitude of the net angular change @xmath166 had to be greater than @xmath167 .
in addition we specified that the magnitude of the net average azimuthal rotation rate @xmath168 had to be greater than @xmath169 where @xmath126 is the lsc turnover time and @xmath170 is the duration of the reorientation ( we refer to @xcite for further details ) . using these criteria
, we found that the number of reorientation events @xmath171 at constant @xmath103 decreased rapidly with increasing @xmath105 .
these results are shown in fig .
[ fig : n ] .
it is worth noting that nearly all of these events are rotations of the lsc and very few involved a cessation of the circulation
. a least - squares fit of the gaussian function @xmath172 \label{eq : n_r}\ ] ] to the data yielded @xmath173 events per hour , @xmath174 rad , and @xmath175 rad .
it is shown by the solid line in the figure .
we note that the distribution function is not centered on @xmath24 .
the displacement of the center by about 9 mrad is much more than the probable error of @xmath0 .
we believe that it is caused by the effect of the coriolis force on the lsc that will be discussed in more detail elsewhere [ @xcite ] .
we saw from fig . [ fig : delta ] that the increase of @xmath21 with @xmath0 led to an increase of the amplitude @xmath19 of the azimuthal temperature variation at the horizontal mid - plane .
an additional question is whether the tilt - angle effect on this system has an asymmetry between the top and bottom that would lead to a change of the mean center temperature @xmath59 ( see eq .
[ eq : t_i ] ) .
@xcite report such an effect for their @xmath44 sample . for a boussinesq sample with @xmath24
we expect that @xmath176 with @xmath177 ( @xmath79 and @xmath78 are the top and bottom temperatures respectively ) , or equivalently that @xmath178 is equal to @xmath179 .
a difference between @xmath180 and @xmath181 will occur when the fluid properties have a significant temperature dependence [ @xcite ] , _ i.e. _ when there are significant deviations from the boussinesq approximation .
for the sequence of measurements with the large apparatus and @xmath103 as a function of @xmath0 the mean value of @xmath182 was @xmath183c and @xmath184 was 0.97@xmath72c , indicating a significant non - boussinesq effect . in fig .
[ fig : nob]a we show @xmath181 and @xmath180 as a function of @xmath0 .
one sees that increasing @xmath0 does not have a significant effect for our @xmath47 sample .
this is shown with greater resolution in fig .
[ fig : nob]b where @xmath185 is shown .
we believe that the small variation , over a range of about @xmath186c , is within possible systematic experimental errors and consistent with the absence of a tilt - angle effect .
in this paper we reported on an experimental investigation of the influence on turbulent convection of a small tilt angle @xmath0 relative to gravity of the axes of two cylindrical rayleigh - bnard samples . the aspect ratios were @xmath48 . where there was overlap , there were significant differences between our results and those obtained by @xcite for a @xmath44 sample .
we found our system to establish a statistically stationary state quickly , within a couple of hours , after a rayleigh - number change whereas @xcite found long transients that they attributed to changes of the lsc structure .
we found a very small depression of the nusselt number @xmath15 with increasing @xmath0 , by about 4% per radian at small @xmath0 .
@xcite found a decrease by 200% per radian for their sample .
in contrast to the very small effect of @xmath0 on @xmath15 , we found an increase of the reynolds number @xmath21 by about 180% per radian for small @xmath0 .
the small effect on @xmath15 in the presence of this large change of @xmath21 indicates that the heat transport does not depend strongly on the speed of the lsc sweeping over the boundary layers .
instead , @xmath15 must be determined by instability mechanisms of the boundary layers , and the associated efficiency of the ejection of hot ( cold ) volumes ( so - called plumes " ) of fluid from the bottom ( top ) boundary layer .
it is interesting to note that the strong dependence of @xmath21 on @xmath0 in the presence of only a very weak dependence of @xmath15 on @xmath0 can be accommodated quite well within the model of @xcite .
the reynolds number can be changed by introducing a @xmath0-dependence of the parameter @xmath187 in their eqs .
( 4 ) and ( 6 ) .
as pointed out by them , a change of @xmath188 has no influence on the predicted value for @xmath15 .
we also measured the frequency of rapid lsc reorientations that are known to occur for @xmath24 .
we found that such events are strongly suppressed by a finite @xmath0 .
even a mild breaking or the rotational invariance , corresponding to @xmath189 , suppresses re - orientations almost completely .
we are grateful to siegfried grossmann and detlef lohse for fruitful exchanges .
this work was supported by the united states department of energy through grant de - fg02 - 03er46080 .
2002 rayleigh - bnard cells : influence of the plates properties _ advances in turbulence ix , proceedings of the ninth european turbulence conference _ , edited by i.p .
castro and p.e .
hancock ( cimne , barcelona ) . | we report experimental results for the influence of a tilt angle @xmath0 relative to gravity on turbulent rayleigh - bnard convection of cylindrical samples .
the measurements were made at rayleigh numbers @xmath1 up to @xmath2 with two samples of height @xmath3 equal to the diameter @xmath4 ( aspect ratio @xmath5 ) , one with @xmath6 m ( the large " sample ) and the other with @xmath7 m ( the medium " sample ) . the fluid was water with a prandtl number @xmath8 .
in contrast to the experiences reported by @xcite for a similar sample but with @xmath9 ( @xmath10 and @xmath11 m ) , we found no long relaxation times . for @xmath12 we measured the nusselt number @xmath13 as a function of @xmath0 and obtained a small @xmath0 dependence given by @xmath14 $ ] when @xmath0 is in radian .
this depression of @xmath15 is about a factor of 50 smaller than the result found by @xcite for their @xmath9 sample .
we measured side - wall temperatures at eight equally spaced azimuthal locations on the horizontal mid - plane of the sample and used them to obtain cross - correlation functions between opposite azimuthal locations .
the correlation functions had gaussian peaks centered about @xmath16 that corresponded to half a turn - over time of the large - scale circulation ( lsc ) and yielded reynolds numbers @xmath17 of the lsc . for the large sample and @xmath12 we found
@xmath18 $ ] .
similar results were obtained also from the auto - correlation functions of individual thermometers .
these results are consistent with measurements of the amplitude @xmath19 of the azimuthal side - wall temperature - variation at the mid - plane that gave @xmath20 $ ] for the same @xmath1 .
an important conclusion is that the increase of the speed ( i.e. of @xmath21 ) with @xmath0 of the lsc does not significantly influence the heat transport .
thus the heat transport must be determined primarily by the instability mechanism operative in the boundary layers , rather than by the rate at which plumes " are carried away by the lsc .
this mechanism apparently is independent of @xmath0 . over the range @xmath22 the enhancement of @xmath17 at constant @xmath0 due to the tilt
could be described by a power law of @xmath1 with an exponent of @xmath23 , consistent with a simple model that balances the additional buoyancy due to the tilt angle by the shear stress across the boundary layers .
even a small tilt angle dramatically suppressed the azimuthal meandering and the sudden reorientations characteristic of the lsc in a sample with @xmath24 . for large
@xmath1 the azimuthal mean of the temperature at the horizontal mid - plane differed significantly form the average of the top- and bottom - plate temperatures due to non - boussinesq effects , but within our resolution was independent of @xmath0 . |
the idea of studying `` network motifs '' in biological systems , namely simple , frequently occurring , structures from which some properties of the system may be inferred , has attracted considerable recent interest ( for example @xcite ) . the reason is quite natural : systems biology proceeds dually by elucidating the functioning of subsystems while at the same time attempting to fit together these pieces to explain large - scale behaviours in biological systems . at the heart of this study is a need to know how behaviours in a dynamical system are affected by embedding this system in some larger system in well - defined ways .
some rigorous results in this area come from the literature on hyperbolicity and structural stability ( @xcite , for example ) , and on monotone control systems ( @xcite , for example ) .
a closely related type of question is when a behaviour known to occur in some member of a _ family _ of dynamical systems @xmath0 can be guaranteed to occur in some member of a new family @xmath1 related to @xmath0 in some natural sense .
our goal here is understanding the inheritance of _ multiple positive nondegenerate equilibria _
( mpne , definition [ defmpne ] below ) and _ multiple positive linearly stable equilibria _ ( mpse , definition [ defmpse ] below ) in chemical reaction networks ( crns ) which form the main component of most biological models .
the literature on multiple equilibria in crns is extensive and we do not attempt to provide a bibliography ; the recent papers @xcite , @xcite , and @xcite provide some references to key strands of this work . but most work , including our own , has focussed on conditions for _ precluding _ multistationarity ; in some ways the opposite of the question treated here .
an exception is the paper by joshi and shiu @xcite , which inspired this paper .
the authors treated questions of inheritance of multistationarity primarily , but not exclusively , focussed on _ fully open _ crns ( definition [ deffullyopen ] below ) .
they highlighted that in studying multistationarity ( and potentially other nontrivial behaviours in crns ) , we need to identify a relevant partial order on the set of all crns , and then search for minimal elements w.r.t .
this partial order , which they termed _
atoms_. this partial order @xmath2 should be such that if some crn @xmath3 displays the behaviour of interest , and @xmath4 , then @xmath5 must display this behaviour . in the context of multistationarity in fully open crns , and with certain kinetic assumptions , there is a nice result : if we identify crns with their natural digraphs ( see definition [ defpetrinet ] below ) , then mpne and mpse are inherited in the induced subnetwork partial order ( see definition [ defiso ] ) .
this result ( corollary [ coropeninduced ] below ) was largely known to joshi and shiu ( corollary 4.6 of @xcite ) .
in fact , the results here allow considerably stronger claims about the inheritance of mpne and mpse in fully open networks ( see remark [ remfobest ] ) . results about fully open crns do not generalise easily to arbitrary crns : a crn which forbids multistationarity may have induced subnetworks which permit multistationarity . building larger crns from smaller ones in natural ways such as adding reactions can quite generally both introduce and destroy multistationarity .
nevertheless , it is possible to formulate several operations which together define a partial order @xmath6 on crns such that if @xmath3 displays mpne ( resp . ,
mpse ) and @xmath7 then @xmath5 displays mpne ( resp . ,
mpse ) , and it is this task which we _ begin _ in this paper . for example , we can add new reactions involving new species while preserving mpne and mpse provided certain conditions are met ( theorem [ thmblockadd ] ) . feliu and wiuf @xcite show that under various circumstances multistationarity survives when a reaction is `` split '' with the introduction of a new intermediate species .
we have a result involving splitting reactions and inserting intermediate complexes ( theorem [ thmintermediates ] ) which reproduces and generalises some elements of their results .
availability of a partial order relevant to mpne or mpse couples naturally with the task of characterising minimal networks admitting these behaviours w.r.t . this partial order .
although minimal networks need not be small , in the sense of having few species , few reactions , or limited numbers of molecules participating in each reaction , a first step is naturally to identify _ small _ minimal subnetworks with the desired behaviour .
we do not undertake this task here , but it was begun for the induced subnetwork partial order in @xcite and continued in the recent preprint @xcite . our proofs all follow roughly the same template , relying heavily on the implicit function theorem , with some insight from differential topology .
we do not need to appeal to degree theory @xcite , or homotopy theory @xcite , or any nontrivial algebra , even though many of the resulting claims are algebraic in the sense of being about the zeros of polynomial equations .
our local approach , though powerful , necessarily has limitations discussed in remark [ remlimit ] and the conclusions .
the paper is laid out as follows . after introducing some basic notions in section [ secbackground ] we present an extended example ( section [ secextended ] ) beginning with a crn which displays mpne and illustrating some of the many ways one might `` build up '' a reaction network , some of which will be proved to preserve mpne and some of which destroy it .
this foreshadows the theorems to follow , while also highlighting their limitations .
we then present in section [ secthms ] the results , some known and some new , but defer the proofs until section [ secproofs ] , after illustrating application of the results on a biologically important crn in section [ secbioexample ] .
a consequence of the analysis in section [ secbioexample ] is previewed in figure [ fig : mapk ] .
finally , we present some discussion and conclusions .
we provide only a brief , informal introduction to crns , and the reader is referred to @xcite for a more thorough discussion in a closely related framework .
@xmath8 refers to the nonnegative orthant in @xmath9 , namely @xmath10 , while @xmath11 refers to the positive orthant , namely @xmath12 . for @xmath13
, @xmath14 will mean that @xmath15 .
a vector is nonnegative ( resp . , positive ) if each component is nonnegative ( resp . ,
positive ) .
consider a set of @xmath16 chemical species @xmath17 with concentrations @xmath18 , and a crn involving @xmath19 reactions between these species .
let the system have _ stoichiometric matrix _
@xmath20 with rank @xmath21 . here
@xmath20 is an @xmath22 real matrix , and @xmath23 is the net change in concentration of species @xmath24 when reaction @xmath25 occurs .
the @xmath25th column of @xmath20 is the _ reaction vector _ for the @xmath25th reaction . in spatially homogeneous , deterministic , continuous time models ,
the evolution of the species concentrations is governed the ode : @xmath26 where @xmath27 is the vector of reaction rates .
we assume that @xmath27 is defined and @xmath28 on @xmath11 .
it is natural to put some further minimal assumptions on @xmath27 , so we always assume that @xmath27 belongs to the class of _ positive general kinetics _ defined formally in @xcite : in brief , the rate of an irreversible reaction ( i ) is positive on the positive orthant ; ( ii ) depends only on the concentrations of its substrates ; and ( iii ) increases strictly with concentration of each substrate on the positive orthant .
restrictions on the functions describing the rate of a reversible reaction are easily derived from the rules for irreversible reactions .
positive general kinetics includes mass action ( ma ) kinetics as a special case , and in fact all our inheritance results are consistent with ma kinetics .
in other words , we choose ma kinetics for any added reactions , and where a reaction is modified in some way , if the original reaction had ma kinetics , then so does the new one . throughout the text , we include remarks about other classes of kinetics for which the results hold often requiring only minor , formal modification of the proofs .
a complex is a formal linear combination of species .
if the species of a crn are @xmath17 , and @xmath29 is a nonnegative integer vector in @xmath9 , we write @xmath30 for the complex @xmath31 .
the zero complex @xmath32 will be denoted @xmath33 .
[ defpetrinet ] consider a crn @xmath3 , each of whose reactions is considered to be irreversible ( namely reversible reactions are considered as pairs of reactions ) .
the _ petri net graph ( pn graph ) _ of @xmath3 , denoted @xmath34 , and also known as the species - reaction petri net of @xmath3 @xcite , is an edge - weighted bipartite digraph .
its two vertex sets @xmath35 ( species vertices ) and @xmath36 ( reaction vertices ) correspond to the species and the reactions of @xmath3 , and given @xmath37 and @xmath38 , there exists an arc @xmath39 ( resp . ,
@xmath40 ) with weight @xmath41 if and only if the species corresponding to @xmath42 occurs with stoichiometry @xmath41 as a substrate ( resp . , product ) in reaction corresponding to @xmath27 .
arcs are labelled with the stoichiometry with which the species figures in the reaction ( with the convention that labels of @xmath43 are omitted ) .
petri net graphs provide one natural way of visualising and analysing crns .
they are closely related to various other bipartite graphs associated with crns , particularly sr and dsr graphs @xcite .
we illustrate below the pn graph of @xmath44 , @xmath45 : ( 1,0.5 ) circle ( 1.5pt ) ; ( 1,-0.5 ) circle ( 1.5pt ) ; ( 3,0.5 ) circle ( 1.5pt ) ; ( 3,-0.5 ) circle ( 1.5pt ) ; at ( 0,0 ) @xmath33 ; at ( 2,0 ) @xmath46 ; at ( 4,0 ) @xmath47 ; ( 0.1,0.1 ) .. controls ( 0.3,0.3 ) and ( 0.5,0.5 ) .. ( 0.9,0.5 ) ; ( 1.85,0.15 ) .. controls ( 1.7,0.3 ) and ( 1.5,0.5 ) .. ( 1.1,0.5 ) ; ( 1.85,-0.15 ) .. controls ( 1.7,-0.3 ) and ( 1.5,-0.5 ) .. ( 1.1,-0.5 ) ; ( 0.1,-0.1 ) .. controls ( 0.3,-0.3 ) and ( 0.5,-0.5 ) .. ( 0.9,-0.5 ) ; ( 2.15,0.15 ) .. controls ( 2.3,0.3 ) and ( 2.5,0.5 ) .. ( 2.9,0.5 ) ; ( 3.85,0.15 ) .. controls ( 3.7,0.3 ) and ( 3.5,0.5 ) .. ( 3.1,0.5 ) ; ( 2.15,-0.15 ) .. controls ( 2.3,-0.3 ) and ( 2.5,-0.5 ) .. ( 2.9,-0.5 ) ; ( 3.85,-0.15 ) .. controls ( 3.7,-0.3 ) and ( 3.5,-0.5 ) .. ( 3.1,-0.5 ) ; ( 3.85,0.05 ) .. controls ( 3.5 , 0.05 ) and ( 3.3 , 0.2 ) .. ( 3.07,0.43 ) ; at ( 3.35,0.05 ) @xmath48 ; at ( 3.6,0.5 ) @xmath49 ; [ defiso ] crns @xmath50 and @xmath51 are _ isomorphic _ if @xmath52 and @xmath53 are isomorphic as edge - labelled digraphs , via an isomorphism which preserves the given bipartition into species and reaction vertices .
this is equivalent to ( independently ) relabelling species and reactions .
an _ induced subnetwork _ of a crn @xmath3 is any crn whose pn graph is isomorphic to a vertex - induced subgraph of @xmath34 , namely can be obtained by deleting some subset of vertices and all their incident arcs from @xmath34 .
if we say that `` @xmath50 is an induced subnetwork of @xmath51 '' , we mean that @xmath50 is isomorphic to an induced subnetwork of @xmath51 .
it is straightforward that the relationship `` is an induced subnetwork '' corresponds to a partial order on the set of crns .
although it is not a mathematical requirement , it is quite natural to remove reactions with no species ( isolated reaction vertices ) , species which participate in no reactions ( isolated species vertices ) , and copies of reactions , from an induced subnetwork . if we were to add these operations into the definition of an induced subnetwork we obtain precisely the definition of `` embedded networks '' in definition 2.2 of @xcite .
[ deffullyopen ] a crn involving species @xmath54 is _ fully open _ if it includes all the inflow - outflow reactions @xmath55 ( @xmath56 ) .
the _ fully open extension _ of a crn @xmath3 is created by adjoining to @xmath3 any such reactions which are absent from @xmath3 .
@xmath57 is referred to as the _ stoichiometric subspace _ of the crn in ( [ gencrn ] ) .
the nonempty intersection of a coset of @xmath57 with the nonnegative ( resp . , positive )
orthant is a _ stoichiometry class _ ( resp .
, _ positive stoichiometry class _ ) of the crn .
stoichiometry classes are easily shown to be invariant for ( [ gencrn ] ) under very general assumptions .
we say that @xmath58 are _ compatible _ if @xmath59 .
the crn evolving according to ( [ gencrn ] ) has _ multiple positive equilibria _ ( mpe )
if there exist @xmath60 s.t .
@xmath59 and @xmath61 . note that it is always implicit in our use of the term `` multiple equilibria '' that the equilibria are compatible .
* nondegeneracy and stability . *
we say that a point @xmath62 is _ nondegenerate _ for ( [ gencrn ] ) if the jacobian determinant of the restricted vector field @xmath63 is nonzero at @xmath62 ( see also definition 4 in @xcite ) .
nondegeneracy is most naturally expressed via reduced jacobian matrices and the reduced determinant of @xmath64 , defined as follows .
[ notationreddet ] consider a differentiable function @xmath65 , and suppose that @xmath66 , where @xmath67 is some @xmath68-dimensonal linear subspace of @xmath9 .
let @xmath69 be some matrix whose columns are a basis for @xmath67 , so there is a unique differentiable function @xmath70 s.t .
( @xmath72 is the image of @xmath73 under @xmath74 in the local coordinates on @xmath67 given by the columns of @xmath69 . )
we term @xmath75 the _ reduced jacobian matrix _ of @xmath74 w.r.t .
a different choice of basis for @xmath67 , written as the columns of a matrix @xmath76 , leads to a reduced jacobian matrix @xmath77 similar to @xmath78 .
when interested only in the spectral properties of such reduced jacobian matrices , we refer to the linear operator they represent as @xmath79 .
we refer to @xmath80 as the _ reduced determinant _ of @xmath74 w.r.t .
@xmath67 .
[ defreddet ] a point @xmath62 is _ nondegenerate _ for ( [ gencrn ] ) if the reduced determinant of @xmath81 w.r.t .
@xmath57 is nonzero at @xmath62 , namely @xmath82 .
@xmath83 can be computed in various ways whose equivalence is shown in @xcite .
most simply , @xmath83 is the sum of @xmath84 principal minors of @xmath85 where @xmath86 , namely upto a sign change , the leading nonzero coefficient in the characteristic polynomial of @xmath85 .
alternatively , following definition [ notationreddet ] , we may choose @xmath87 to be any matrix whose columns are a basis for @xmath57 and define @xmath88 via @xmath89 .
then @xmath90 .
[ defmpne ] a crn of the form ( [ gencrn ] ) with kinetics in some fixed class ( such as positive general kinetics or mass action kinetics ) admits _ multiple , positive , nondegenerate equilibria _ ( mpne )
if there exists _ some _ choice of allowed kinetics s.t .
there exist a pair of nondegenerate equilibria on _ some _ positive stoichiometry class , namely there exist @xmath91 satisfying @xmath59 with @xmath92 and @xmath93 .
we say that a square matrix @xmath94 is _ hurwitz stable _ if its eigenvalues all lie in @xmath95 , the ( open ) left half plane of @xmath96 .
[ defmpse ] a crn of the form ( [ gencrn ] ) with kinetics in some fixed class admits _ multiple positive linearly stable equilibria _ ( mpse )
if there exists _ some _ choice of allowed kinetics s.t .
it has two positive , compatible equilibria , both linearly stable w.r.t .
their stoichiometry class , namely @xmath91 satisfying @xmath59 with @xmath97 and @xmath98 both hurwitz stable .
a crn which admits mpse admits mpne by definition .
* inheritance of mpne / mpse . * given a crn @xmath3 displaying mpne ( resp . , mpse ) , we will present a number of ways in which we can modify @xmath3 to create a new crn @xmath5 in such a way that @xmath5 displays mpne ( resp . ,
this will generally involve adding more reactions and/or more species , or modifying reactions in some way .
a similar project could be undertaken for other dynamical behaviours , in particular for crns which admit hyperbolic periodic orbits , and we remark on this after certain proofs . at the heart of our proofs
is the implicit function theorem ( ift ) which , for our purposes here , takes the form : [ lemift ] let @xmath99 be open and @xmath100 be @xmath28 .
suppose @xmath101 for some @xmath102 and the jacobian matrix @xmath103 ( namely with respect to the first variables ) is nonsingular .
then there exist @xmath104 , @xmath105 both open with @xmath106 , and a @xmath28 function @xmath107 satisfying @xmath108 , and @xmath109 namely the set of zeros of @xmath74 in @xmath110 is precisely the graph of @xmath111 .
see , for example , theorem 9.28 in @xcite .
application of the ift will frequently take the following form : [ corift]let @xmath112 be open and let @xmath113 be a @xmath68-dimensional linear subspace of @xmath9 .
let @xmath114 and consider a @xmath28 function @xmath115 s.t .
for each fixed @xmath116 , @xmath117 satisfies @xmath118 .
let @xmath119 be a nondegenerate zero of @xmath120 w.r.t .
@xmath67 , namely , @xmath121 and @xmath122 . then there exists @xmath123 and a @xmath28 curve @xmath124 with @xmath125 and @xmath126 for @xmath127 .
moreover , @xmath128 is an isolated zero of @xmath129 on @xmath130 .
as in definition [ notationreddet ] , let @xmath69 be a matrix whose columns are a basis for @xmath67 , and define @xmath131 via @xmath132 .
let @xmath133 . note that @xmath47 is open and includes the origin .
define @xmath134 by @xmath135 .
note that ( fixing @xmath116 ) @xmath136 ( where @xmath137 ) , namely the reduced jacobian matrix of @xmath129 w.r.t .
@xmath69 is just the jacobian matrix of @xmath138 w.r.t .
its first argument .
as @xmath62 is a nondegenerate zero of @xmath120 , we have @xmath139 and @xmath140 is nonsingular .
we may then apply the ift to @xmath138 giving an open neighbourhood @xmath141 of @xmath33 in @xmath142 , some @xmath143 , and a function @xmath144 s.t .
@xmath145 the result follows with @xmath146 .
[ remlimit ] the results here are all * local * , and we never use global algebraic methods . a local approach ,
although powerful , necessarily has limitations : we will see by example in the following section that there are situations where the capacity for mpne is inherited , but we are unlikely to see this via a local argument .
to motivate the results to follow and highlight their limitations we present an extended example beginning with a crn @xmath147 displaying mpne and then performing various modifications to create crns @xmath50 to @xmath148 .
each modification can either be proved to preserve mpne using a result in the paper , or highlights some point about the scope of these theorems .
@xmath147 appears in remark 4.4 of joshi and shiu s paper @xcite . *
base reaction network . *
we begin with the following crn on two species : @xmath149 if @xmath147 is given mass action kinetics with both rate constants set at @xmath43 , we get the odes @xmath150 .
positive equilibria occur when @xmath151 .
the stoichiometric matrix has rank @xmath43 and the stoichiometry classes of @xmath147 satisfy @xmath152 for some constants @xmath153 .
the system has reduced determinant @xmath154 ( the trace of the jacobian matrix ) .
it is easy to confirm by direct calculation that any stoichiometry class with @xmath155 contains two positive equilibria at @xmath156 , @xmath157 , where @xmath158 .
the reduced determinant is @xmath159 and @xmath160 at @xmath62 and @xmath161 respectively , and hence , as @xmath162 implies @xmath163 and @xmath164 , @xmath62 and @xmath161 are nondegenerate .
thus @xmath147 admits mpne . that @xmath147 has mpne
can also be shown as an immediate application of theorem 4.5 in @xcite .
@xmath165 * modification 1 : adding new linearly dependent reactions preserves mpne . * consider adding to @xmath147 some new reaction with mass action kinetics and with reaction vector which is a linear combination of existing reaction vectors , so that the stoichiometric subspace does not change .
for example we might enlarge @xmath147 with the reaction @xmath166 to get @xmath167 intuition says that provided we can make the rate of the new reaction small on any compact set , its addition represents only a small perturbation to the vector field on any compact region of some particular stoichiometry class , and hence mpne should occur for @xmath50 .
this is easily formalised as a general result ( theorem [ thmnewdepreac ] ) .
the following three examples show that adding new reactions which are not linearly dependent on the old ones may have different effects on mpne . *
modification 2 : adding new independent reactions may destroy mpne .
* consider now adding to @xmath147 the reversible reaction @xmath168 with mass action kinetics , to get @xmath169 the unique stoichiometry class of this system is now @xmath170 , and the unique positive equilibrium of this system is at @xmath171 .
regardless of how we choose the rate constants , mpne has been destroyed .
thus in general the capacity for mpne is not inherited by supernetworks of a crn . * modification 3 : adding new independent reactions may allow multiple degenerate equilibria . *
consider adding to @xmath147 the reaction @xmath172 with mass action kinetics , to get @xmath173 the unique stoichiometry class of @xmath174 is @xmath170 and the reduced determinant is easily computed to be @xmath175 . if @xmath176 , this system has no equilibria at all . on the other hand
if @xmath177 , then the set of equilibria consists of solutions to @xmath178 , and all these equilibria are degenerate .
thus the capacity for mpne is destroyed , even though @xmath174 still allows multiple positive _ degenerate _
equilibria for special choices of rate constants .
* modification 4 : adding new independent reactions may preserve mpne . *
consider adding to @xmath147 three irreversible reactions @xmath179 , @xmath180 and @xmath181 , and choosing mass action rate constants as indicated : @xmath182 the unique stoichiometry class of @xmath183 is @xmath170 .
positive equilibria are solutions to @xmath184 .
the hyperbola and circle have two positive intersections at @xmath185 for @xmath186 and @xmath187 , and these two points correspond to nondegenerate equilibria .
thus the capacity for mpne is preserved
. however , there is no obvious _ local _ argument for why mpne should be preserved in this case : the mpne constructed for the new system are not in any sense derived as perturbations of original mpne .
* modification 5 : adding inflows and outflows of all species preserves mpne . *
suppose we add to @xmath147 inflows and outflows of all species to get the fully open extension of @xmath147 ( definition [ deffullyopen ] ) : @xmath188 the stoichiometric subspace becomes all of @xmath189 .
this modification is well known to preserve mpne .
a variety of proofs are known and a very simple ift - based argument is possible ( theorem [ thmopenextension ] ) .
@xmath165 * modification 6 : adding a trivial species preserves mpne . *
suppose we add a new species @xmath190 into the reactions of @xmath147 in a rather trivial way : the species always figures with the same stoichiometry on both sides of any reaction in which it participates .
we might get for example : @xmath191 the stoichiometry classes of @xmath192 are still one dimensional and are parallel to the @xmath193 plane ( namely , @xmath194 is constant ) .
intuition suggests that mpne should be preserved , and this turns out to be the case .
although this result ( theorem [ thmtrivial ] ) is very easy to prove , it is quite important as a step in building crns displaying mpne from smaller ones .
* modification 7 : adding new species may destroy mpne . *
suppose we add a new species into the reactions of @xmath147 to get , for example , @xmath195 in this case , stoichiometry classes of @xmath196 remain one dimensional and are defined by the restrictions @xmath197 , @xmath198 , and positive equilibria occur when @xmath199 where @xmath200 is some positive constant ( a ratio of the rate constants ) .
a little substitution gives that @xmath201-values of positive equilibria occur when @xmath202 , while @xmath73 and @xmath194 values are then uniquely determined .
as this quadratic has exactly one positive root ( for any positive values of @xmath203 and @xmath200 ) , the capacity for mpne is destroyed .
thus , in general , adding new species into reactions may destroy mpne .
* modification 8 : adding new species may preserve mpne .
* suppose , instead , we add the new species into existing reactions to get : @xmath204 stoichiometry classes of @xmath205 remain one dimensional and are defined by the restrictions @xmath197 , @xmath198 , and positive equilibria occur when @xmath206 where @xmath200 is some positive constant ( a ratio of the rate constants ) . choosing @xmath207 and @xmath208 , a little substitution gives that @xmath201-values of positive equilibria occur when @xmath209 .
we can check , for example , that @xmath210 , and @xmath211 are positive , stoichiometrically compatible equilibria of @xmath205 .
thus the capacity for mpne is preserved in this case .
it remains unclear whether there is a local argument for why this should be the case .
@xmath165 * modification 9 : adding new reactions involving new species often preserves mpne . *
consider adding to @xmath147 a single reversible reaction @xmath212 involving a new species @xmath190 to get : @xmath213 the evolution now occurs in @xmath214 , and the stoichiometric subspace of @xmath215 is now @xmath216-dimensional .
it seems plausible ( though by no means obvious ) that choosing the rate constants of the new reaction appropriately should lead to mpne in the new crn , and indeed this turns out to be the case .
this is a simple case of a much more general result ( theorem [ thmblockadd ] ) allowing the addition of several new reactions involving several new species .
we remark that theorem [ thmblockadd ] includes a nondegeneracy condition which in this case is equivalent to the requirement that the new species must not occur with the same stoichiometry on both sides of the new reaction .
the reader may easily confirm , for example , that adding @xmath217 to @xmath147 , results in a crn with a unique equilibrium on each stoichiometry class for all rate constants .
* modification 10 : adding a new species into reactions while also adding its inflow and outflow preserves mpne .
* suppose we add into some reactions of @xmath147 a new species @xmath190 ( in any way ) while also adding its inflow and outflow @xmath218 .
we may get , for example @xmath219 intuition suggests that if the rates of the added inflow - outflow reaction are large enough then this reaction effectively holds @xmath190 constant and thus the net effect is only to modify the rates of the reactions in which @xmath190 participates .
mpne should survive under mild assumptions on the kinetics .
indeed this is the case , and it is easy to make the intuitive argument precise ( theorem [ thmnewwithopen ] ) .
@xmath165 * modification 11 : `` splitting a reaction '' and inserting a complex involving a new species often preserves mpne .
* suppose that we modify an existing irreversible reaction of @xmath147 by adding a new complex containing some new species @xmath190 , as an intermediate .
for example we might get : @xmath220 the evolution now occurs in @xmath214 , and the stoichiometric subspace is now @xmath216-dimensional .
it seems plausible that choosing the rate constants of the new reactions appropriately should lead to mpne in the new crn , and indeed this turns out to be the case , as a consequence of one of our general results ( theorem [ thmintermediates ] ) .
here , as in modification 9 , the general result allows adding several new species in several intermediate complexes provided a nondegeneracy condition is met .
* modification 12 : splitting a reaction and inserting a complex involving no new species may destroy mpne .
* suppose that we modify an existing irreversible reaction of @xmath147 by adding a new complex comprised of existing species as an intermediate .
for example we might get : @xmath221 the evolution occurs in @xmath189 , and the stoichiometric subspace is now @xmath216-dimensional .
equilibria occur when @xmath222 , and we easily compute that the only positive solution to these equations is @xmath223{\frac{k_3 ^ 2}{k_1k_2 } } , y=\sqrt[3]{\frac{k_3k_2}{k_1 ^ 2}}$ ] .
thus the capacity for mpne is destroyed by this modification . 1 .
if we add new `` dependent '' reactions , as in @xmath224 , then we can quite generally expect mpne and mpse to survive ( theorem [ thmnewdepreac ] ) .
this is also the subject of theorem 3.1 in @xcite .
if we add new `` independent '' reactions on existing species , as in modifications 25 , then the consequences are unclear .
mpne and mpse are preserved in the important and well known special case of taking the `` fully open extension '' of the original crn as in @xmath225 ( theorem [ thmopenextension ] ) .
but in general , such modifications may : * lead to the destruction of mpne as in @xmath226 . *
destroy mpne while still allowing multiple _ degenerate _ equilibria as in @xmath227 .
* preserve mpne as in @xmath228 .
if we add species into reactions , without adding new reactions , then mpne and mpse automatically survive if the added species figure trivially in reactions ( theorem [ thmtrivial ] ) . on the other hand
, @xmath229 shows that in general we can not hope for mpne to survive , and where it does , as in @xmath230 , there may be no obvious local argument for this .
if we add new species and new reactions to a crn , then mpne and mpse survive in the important special case where we add an inflow and outflow reaction for each new species , illustrated in @xmath231 ( theorem [ thmnewwithopen ] below , and see also theorem 4.2 in @xcite ) .
if we add new reversible reactions involving some new species , without adding the new species into any existing reactions ( as in @xmath232 ) , then provided certain conditions are met mpne and mpse survive ( theorem [ thmblockadd ] ) .
finally , @xmath233 and @xmath148 provide examples apparently of a rather different kind , where we modify a network by adding intermediate complexes into reactions , effectively both deleting and adding reactions .
we will see that actually this modification is formally similar to adding new reactions . under certain conditions ,
this preserves mpne and mpse , as in @xmath233 ( theorem [ thmintermediates ] ) .
the addition of intermediate complexes is also the subject of @xcite , and we comment later on the relationship with results in this paper .
we gather together the results about the inheritance of mpne , presenting the proofs later .
theorems [ thmnewdepreac]-[thmnewwithopen ] are easy and some elements of these results are well - known .
theorems [ thmblockadd ] and [ thmintermediates ] are harder . in each case
, @xmath3 is an unknown crn which admits mpne , and we assume that the kinetics is fixed such that mpne occurs . @xmath62 and @xmath161 are a pair of positive , compatible , nondegenerate equilibria of @xmath3 .
the species of @xmath3 are @xmath17 , with corresponding concentration vector @xmath234 .
the stoichiometric matrix of @xmath3 , @xmath20 , is an @xmath22 matrix with rank @xmath21 , and the reaction rates of @xmath3 are fixed at @xmath235 .
columns of the matrix @xmath87 are a basis for @xmath236 , and @xmath88 is defined via @xmath89 ( see definition [ defreddet ] ) .
@xmath237 refers to the coset of @xmath67 containing @xmath62 , and by assumption @xmath238 .
@xmath5 is created from @xmath3 by adding species and/or reactions , and/or modifying reactions .
any new species are termed @xmath47 with concentration vector @xmath201 . in every case the new kinetics
is chosen in such a way that if @xmath3 has ma kinetics , then so does @xmath5 , while if @xmath3 has positive general kinetics , then so does @xmath5 .
the proofs make clear how the kinetic assumptions can be loosened , and we present remarks after the proof of each result on this . theorem [ thmnewdepreac ] is closely related to theorem 3.1 in @xcite .
[ thmnewdepreac ] suppose we create @xmath5 from @xmath3 , by adding to @xmath3 a new irreversible reaction with reaction vector @xmath239 ( namely , @xmath240 is a linear combination of reaction vectors of @xmath3 ) .
then @xmath5 admits mpne .
if @xmath3 admits mpse , then @xmath5 admits mpse .
[ newdeprev ] from theorem [ thmnewdepreac ] , we could add the reverse of any existing reaction to @xmath3 as this would be a dependent reaction .
equally , we could add a new _ reversible _ dependent reaction to @xmath3 and preserve mpne , as this amounts to sequentially adding two new irreversible reactions to @xmath3 .
the claim about mpne in theorem [ thmopenextension ] is well known .
it appears as theorem 2 in @xcite proved using an approach similar to the one here , and as lemma b1 in @xcite proved using techniques involving the invariance of brouwer degree , and closely related to the proof of proposition 1 in @xcite .
[ thmopenextension ] suppose we create @xmath5 from @xmath3 , by adding to @xmath3 all the reactions @xmath55 for each @xmath24 in @xmath241 .
then @xmath5 admits mpne .
if @xmath3 admits mpse , then @xmath5 admits mpse .
the following theorem is very easy to prove , and yet surprisingly important in applications when combined with other results in this paper .
[ thmtrivial ] suppose we create @xmath5 from @xmath3 , by adding into the reactions of @xmath3 a new species @xmath47 which occurs with the same stoichiometry on both sides of each reaction in which it participates .
then @xmath5 admits mpne . if @xmath3 admits mpse , then @xmath5 admits mpse .
the following theorem is closely related to theorem 4.2 in @xcite : [ thmnewwithopen ] suppose we create @xmath5 from @xmath3 , by adding into the reactions of @xmath3 the new species @xmath47 in an arbitrary way , while also adding the new reaction @xmath242
. then @xmath5 admits mpne .
if @xmath3 admits mpse , then @xmath5 admits mpse .
theorem [ thmnewwithopen ] , combined with theorem [ thmnewdepreac ] allows us to deduce an important corollary whose claim about mpne appears , although with different terminology , in @xcite .
[ coropeninduced ] let @xmath50 and @xmath51 be fully open crns with @xmath50 an induced subnetwork of @xmath51 ( definition [ defiso ] ) . if @xmath50 admits mpne then @xmath51 admits mpne .
if @xmath50 admits mpse then @xmath51 admits mpse .
we can construct @xmath53 from @xmath52 via a sequence of steps as follows : ( i ) first , for each absent species ( if any ) we add the species vertex , its inflow and outflow reaction vertices , and all missing arcs ( corresponding to theorem [ thmnewwithopen ] ) ; ( ii ) then , for each remaining absent reaction ( if any ) we add the reaction vertex and all its incident arcs ( corresponding to theorem [ thmnewdepreac ] because a fully open crn has stoichiometric subspace which is the whole ambient space , and hence each added reaction is now a dependent reaction ) . if @xmath50 admits mpne ( resp .
, mpse ) , then since each step preserves this property , @xmath51 admits mpne ( resp . , mpse ) .
the claim about mpne in corollary [ coropeninduced ] appears as corollary 4.6 of @xcite .
the claim about mpse is part of theorem 4.5 in the later work @xcite .
however , it appears that our approach allows considerably weaker kinetic assumptions than needed for corollary 4.6 of @xcite .
although the proofs of theorems [ thmnewdepreac ] and [ thmnewwithopen ] on which corollary [ coropeninduced ] depends are presented only for mass action kinetics or positive general kinetics , the results hold for a wide variety of kinetic assumptions as discussed in remarks [ remnewdepreac ] and [ remnewwithopen ] below including , almost immediately from the proofs , any fixed power - law kinetics .
joshi ( theorem 4.13 of @xcite ) showed that the fully open crns admitting mpne , minimal w.r.t .
the induced subnetwork partial order , and with only one non - flow reaction are the fully open extensions of either @xmath243 or @xmath244 , where @xmath245 , respectively @xmath246 and @xmath247
. corollary [ coropeninduced ] can be improved using theorem [ thmintermediates ] below as indicated in remark [ remfobest ] .
the next two theorems are somewhat harder than theorems [ thmnewdepreac ] to [ thmnewwithopen ] , and can be regarded as the main results of this paper .
[ thmblockadd ] suppose we create @xmath5 from @xmath3 , by adding @xmath248 new reversible reactions involving @xmath68 new species with the following nondegeneracy condition : the submatrix of the stoichiometric matrix of @xmath5 corresponding to the new species has rank @xmath249 ( this implies @xmath250 ) .
then @xmath5 admits mpne .
if @xmath3 admits mpse , then @xmath5 admits mpse .
[ thmintermediates ] suppose we create @xmath5 from @xmath3 by replacing each of the @xmath249 reactions : @xmath251 here @xmath47 is a list of @xmath68 new species which enter nondegenerately into @xmath5 in the sense that @xmath252 has rank @xmath249 ( this implies @xmath250 ) .
@xmath253 , @xmath254 and @xmath255 are arbitrary nonnegative vectors and any or all may coincide .
then @xmath5 admits mpne .
if @xmath3 admits mpse , then @xmath5 admits mpse . in order to understand the scope of theorem [ thmintermediates ] , we note first that we may always formally write a single reaction as multiple reactions ; for example we way consider the single reaction @xmath256 as a set of @xmath249 such reactions , each with rate @xmath257 times the original rate .
this clearly does not affect the associated dynamics and can be done while remaining in any reasonable class of kinetics ( most particularly within mass action kinetics , where each split reaction has rate constant which is the original rate constant divided by @xmath249 ) .
thus theorem [ thmintermediates ] may be used to `` split '' fewer than @xmath249 reactions , with some reactions being split multiple times and acquiring multiple intermediates .
since there are no restrictions on @xmath253 and @xmath254 , many ( but not all ) instances of theorem [ thmblockadd ] also follow from theorem [ thmintermediates ] .
for example , adding to a crn a trivial reaction with the same complex on both sides ( with any kinetics ) does not alter the vector field . given a crn with mpne
, we could carry out this step first , and then `` split '' the trivial reaction using theorem [ thmintermediates ] to create a new reversible reaction involving some new species .
the resulting crn would have mpne by theorem [ thmintermediates ] .
theorem [ thmintermediates ] appears to generalise certain aspects of results in @xcite , while being unable to reproduce others .
in particular , we allow the introduction of complexes ( possibly involving existing species ) rather than just new individual species as intermediates ; however we forbid the introduction of intermediates in a degenerate way allowed in @xcite ( although see the remarks in the conclusions ) .
this divergence is not entirely surprising as the techniques in @xcite are global and algebraic , while those here are local and analytical .
theorems [ thmnewdepreac ] to [ thmintermediates ] collectively define a partial order @xmath6 on the set of all crns : @xmath258 if and only if @xmath51 can be obtained from @xmath50 by a sequence of modifications ( possibly empty ) as described in theorems [ thmnewdepreac ] to [ thmintermediates ] .
reflexivity and transitivity are trivial ; antisymmetry is ensured by the fact that each modification either increases the number of species or the number of reactions or both .
so , for example , the crns @xmath50 , @xmath259 , @xmath192 , @xmath215 , @xmath231 and @xmath233 in section [ secextended ] are all @xmath260 @xmath147 .
[ remfobest ] theorem [ thmintermediates ] allows an improvement on corollary [ coropeninduced ] .
given a fully open crn @xmath3 admitting mpne ( resp .
, mpse ) , we may `` split '' some reactions as in theorem [ thmintermediates ] to get a crn @xmath5 admitting mpne ( resp .
@xmath5 will have some new species , say @xmath47 , but the stoichiometric matrix of @xmath5 has full rank ( namely rank equal to the number of species in @xmath5 ) by the rank condition in theorem [ thmintermediates ] .
we may now add inflows and outflows of the species @xmath47 to get the fully open extension @xmath261 of @xmath5 . by theorem [ thmnewdepreac
] @xmath261 admits mpne ( resp . ,
mpse ) since the added reactions were dependent reactions . in general @xmath261 may be a minimal crn admitting mpne ( resp .
, mpse ) in the induced subnetwork partial order , but by construction is not minimal w.r.t .
the partial order @xmath6 here .
for example , according to theorem 4.13 of @xcite , the crn @xmath262 with mass action kinetics admits mpne and consequently , by theorems [ thmintermediates ] and [ thmnewdepreac ] applied in that order , @xmath263 admits mpne .
it is easy to confirm however , either by direct calculation or by applying theorem 4.13 of @xcite and theorem 3.6 of @xcite , that no fully open induced subnetwork of @xmath3 admits mpne . thus @xmath3 is a minimal fully open crn admitting mpne in the induced subnetwork partial order , while it is not minimal w.r.t .
the partial order @xmath6 here .
the following corollary is of importance in the analysis of biological systems , and involves the operation of adding an enzymatic mechanism to a set of reactions .
it will be used a number of times in the analysis of a particular system in section [ secbioexample ] : [ cor : addenz ] let @xmath264 and @xmath265 be new species , not involved in @xmath266 , and let @xmath267 ( @xmath268 ) .
suppose we create @xmath269 from @xmath270 by replacing each of the reactions @xmath271 then if @xmath266 admits mpne , so does @xmath269 .
if @xmath266 admits mpse , so does @xmath269 .
we perform successive modifications that preserve the capacity for mpne and mpse : 1 .
we add the trivial species e into all the reactions simultaneously ( an application of theorem [ thmtrivial ] ) : @xmath272 2 .
we add the set of intermediate species @xmath273 ( an application of theorem [ thmintermediates ] ) : @xmath274 3 .
we add the dependent reactions @xmath275
( @xmath249 applications of theorem [ thmnewdepreac ] ) : @xmath276 we could see step ( 2 ) as @xmath249 applications of theorem [ thmintermediates ] or a single application of the theorem . in either case , the nondegeneracy condition of theorem [ thmintermediates ] is easily shown to be met .
this completes the proof .
before turning to the proofs of the theorems , we present an example indicating how the results here intersect with published work on biologically important systems .
the mitogen - activated ( mapk ) cascade is an extensively studied network occurring in various biological processes in eukaryotic cells .
the model of the mapk cascade ( figure [ fig : hffeed](f ) ) was developed by huang and ferrell @xcite , who showed that the cascade may exhibit ultrasensitive behaviour , making it appropriate for switch - like processes like mitogenesis or cell - fate determination .
addition of a negative feedback in the mapk cascade ( figure [ fig : hffeed](g ) ) was demonstrated numerically to cause oscillatory behaviour by kholodenko @xcite .
qiao et al .
@xcite later showed numerically that the huang - ferrell model without the added negative feedback can also exhibit oscillatory behaviour and bistability .
bistability in the mapk cascade has also been observed numerically and discussed in earlier papers @xcite . here
we show how straightforward applications of our results prove the existence of mpne and in fact mpse in the mapk cascade , with or without negative feedback . with mass action
, the mapk cascade with negative feedback ( figure [ fig : hffeed](g ) ) involves 25 species and 36 reactions ; we start by proving that the much simpler network in figure [ fig : hffeed](a ) admits mpse , and then apply a sequence of modifications that preserve mpse to ultimately transform network ( a ) into ( g ) and ( f ) of figure [ fig : hffeed ] . in what follows , networks ( a)(g ) refer to figure [ fig : hffeed ] .
we use the following abbreviations in order to aid readability of the reaction networks : x = mapk , y = mkk , z = mkkk . *
( a ) a minimal multistationary subnetwork . *
the two first phosphorylation steps of x ( namely , mapk ) are induced by y - pp ( namely , mkk - pp ) , but only the first one involves an intermediate species . both dephosphorylation steps are direct and involve no enzyme .
the reactions are as follows : @xmath277 this network involves 5 species and 6 irreversible reactions . by an exact calculation ( appendix [ appbio ] )
one can prove that the network with mass action kinetics admits mpne .
this can also be confirmed using crntoolbox @xcite .
we also confirmed numerically that it admits mpse .
moreover , it can be checked that the network can not be constructed from any simpler network which admits mpne using the modifications described in theorems [ thmnewdepreac ] to [ thmintermediates ] . in this sense
the network is a minimal network admitting mpne with respect to the partial order @xmath6 and we might refer to it as an `` atom of multistationarity '' borrowing the terminology of @xcite . *
( b ) adding enzymes : the double phosphorylation network . *
network ( a ) is a simplified version of the much - studied double phosphorylation - dephosphorylation cycle , whose capacity for bistable behaviour has been both demonstrated by numerical simulations , and shown analytically @xcite .
its reaction network @xmath278 is obtained from that of ( [ eq : neta ] ) by adding enzymatic mechanisms as in corollary [ cor : addenz ] : the two reactions @xmath279 and @xmath280 get replaced by the two chains @xmath281 and @xmath282 respectively .
it follows that ( b ) , with mass action kinetics , admits mpse . *
( c ) cascading double phosphorylations i. * a simple version of the mkk double phosphorylation , not mediated by an enzymatic mechanism , is coupled upstream of the double phosphorylation mechanism of mapk from ( b ) .
the network is composed of the reactions in ( [ eq : netb ] ) , plus @xmath283 adding these two reversible reactions involving the two new species y - p ( mkk - p ) and y ( mkk ) preserves the capacity for mpse of the new network .
this can be regarded as two applications of theorem [ thmblockadd ] , or a single application of the theorem ; the nondegeneracy condition is easily seen to be satisfied . therefore ( c ) , with mass action kinetics , admits mpse . *
( d ) cascading double phosphorylations ii . * here the upstream double phosphorylation in ( c ) is modified to include enzymatic mechanisms .
the reaction network of ( d ) @xmath284 is obtained from that of ( c ) by replacing the four irreversible reactions ( [ eq : netc ] ) by the four chains @xmath285 corollary [ cor : addenz ] guarantees that ( d ) admits mpse . *
( e ) a three layer cascade . * in network
( e ) a third layer is added upstream of cascade ( d ) .
the network is composed of all reactions in ( [ eq : netd ] ) and @xmath286 adding this reversible reaction involving the new species z ( namely , mkkk ) falls immediately under the scope of theorem [ thmblockadd ] . therefore ( e ) , with mass action kinetics , admits mpse . *
( f ) mapk cascade .
* now we modify ( e ) by adding enzymatic mechanisms ( corollary [ cor : addenz ] ) to the reversible reaction ( [ eq : nete ] ) , which is replaced by the two chains @xmath287 and @xmath288 it follows that the new network ( huang - ferrell mapk cascade @xcite , reactions below ) , with mass action kinetics , admits mpse : @xmath289 * ( g ) mapk cascade with negative feedback .
* finally , we add a negative feedback loop from x - pp as in @xcite . in that paper
, the model describes the feedback as noncompetitive inhibition of z phosphorylation , whose mechanism can be unpacked as ( see for example @xcite ) : @xmath290 these reactions , together with those of ( [ eq : netf ] ) make up network ( g ) .
first we add the reversible reactions ( [ eq : nf1 ] ) and ( [ eq : nf2 ] ) , involving the new species e@xmath291x - pp and e@xmath291x - pp z , to the network ( [ eq : netf ] ) . the nondegeneracy condition is easily checked , and by theorem [ thmblockadd ] the resulting network @xmath292 admits mpse .
next , let @xmath293 and @xmath294 denote the reaction vectors of ( [ eq : nf1 ] ) , ( [ eq : nf2 ] ) , ( [ eq : nf3 ] ) and e@xmath291+z @xmath295 @xmath296 , and note that @xmath297 making ( [ eq : nf3 ] ) a dependent reaction for @xmath292 . by theorem [ thmnewdepreac ] ( and remark [ newdeprev ] )
, we may add ( [ eq : nf3 ] ) to @xmath292 , and thus the resulting network ( g ) , with mass action kinetics , admits mpse .
some additional notation and definitions simplify the presentation . given two matrices @xmath94 and @xmath298 with the same dimensions , @xmath299 will refer to the entrywise ( or hadamard ) product of @xmath94 and @xmath298 , namely @xmath300 .
when we apply functions such as @xmath301 and @xmath302 with a vector or matrix argument , we mean entrywise application . similarly , if @xmath234 and @xmath303 , then @xmath304 .
we write @xmath305 for @xmath306 .
we denote the cardinality of a finite set @xmath67 by @xmath307 .
we write @xmath308 for a vector of ones whose length is inferred from the context , and given some parameter @xmath116 , we write @xmath309 for @xmath310 . given @xmath311 and @xmath312 , @xmath313 is an abbreviation for the monomial @xmath314 .
if @xmath94 is an @xmath315 matrix with rows @xmath316 , then @xmath317 means the vector of monomials @xmath318 .
the following examples demonstrate how entrywise and monomial notation greatly abbreviates otherwise lengthy calculations .
let @xmath319 , @xmath320 and @xmath321 .
let @xmath322 refer to the @xmath323 matrix of zeros .
then ( i ) @xmath324 , ( ii ) @xmath325 , and ( iii ) @xmath326 .
suppose @xmath319 , @xmath327 and @xmath328 .
if @xmath329 , then @xmath330 .
suppose @xmath331 , @xmath15 , @xmath332 .
let @xmath333 be defined by @xmath334
. then @xmath335 .
this expression is , of course , very familiar as the jacobian matrix of a rate function of a mass action crn ( @xcite for example . )
given an @xmath323 matrix @xmath94 and ( nonempty ) sets @xmath336 and @xmath337 , define @xmath338 to be the submatrix of @xmath94 with rows from @xmath240 and columns from @xmath339 . if @xmath340 , then @xmath341 : = \mathrm{det}(a(\alpha|\beta))$ ] .
@xmath342 is shorthand for @xmath343 , and @xmath344 $ ] is shorthand for the principal minor @xmath345 $ ] .
[ defcauchybinet ] given @xmath346 and @xmath347 , and any nonempty @xmath348 , @xmath349 with @xmath340 : @xmath350 = \sum_{\substack{\gamma \subseteq \langle m \rangle\\ |\gamma| = |\alpha|}}a[\alpha|\gamma]b[\gamma|\beta].\ ] ] see @xcite , for example .
[ lemschur ] suppose we have a matrix partitioned into four blocks @xmath351 with @xmath94 and @xmath352 square , and @xmath94 nonsingular .
then @xmath353 this useful and well - known formula , based on schur complements , is presented as formula ( i ) on p46 of @xcite .
[ lemhurwitz ] let @xmath354 for some @xmath355 and suppose we have an @xmath116-dependent family of matrices partitioned into four blocks @xmath356 with @xmath94 and @xmath352 square .
suppose further that : ( i ) @xmath357 is defined and hurwitz ; ( ii ) @xmath358 , @xmath359 and @xmath360 are zero matrices ; ( iii ) @xmath361 is defined and hurwitz
. then there exists @xmath362 s.t .
@xmath363 is hurwitz for all @xmath364 .
let @xmath94 be @xmath365 and @xmath352 be @xmath366 .
we calculate @xmath367 by the continuous dependence of eigenvalues of a matrix on its entries , ( [ eqspeca ] ) tells us that @xmath249 eigenvalues of @xmath368 approach the eigenvalues of @xmath369 , and so there exist @xmath370 and @xmath114 s.t . for @xmath371 @xmath249 eigenvalues of @xmath368
lie in @xmath372 , where @xmath373 . on the other hand ,
expanding @xmath374 along the top @xmath16 rows gives : @xmath375 where @xmath376 .
namely , eigenvalues of @xmath363 are the zeros of @xmath377 .
but @xmath378 , so ( since @xmath379 ) the zeros of @xmath380 are the eigenvalues of @xmath381 .
consider any contour @xmath153 in @xmath95 with the spectrum of @xmath381 in its interior .
choose @xmath382 such that @xmath383 implies that @xmath384 on @xmath153 , and @xmath385 .
then by rouch s theorem ( thm . 7.7 in @xcite for example ) , @xmath386 has the same number of zeros in @xmath153 as @xmath380 . thus ( [ eqspecb ] ) tells us that , for @xmath383 , @xmath16 eigenvalues of @xmath363 lie in @xmath153 and hence in @xmath95 ( in fact , it is easy to refine the argument to see that they approach the spectrum of @xmath381 as @xmath387 ) .
the corresponding @xmath16 eigenvalues of @xmath368 lie in @xmath388 . in summary , for @xmath389 , @xmath368 has @xmath249 eigenvalues in @xmath372 and @xmath16 eigenvalues in @xmath388 , and so is hurwitz stable . the same clearly holds for @xmath363 . we begin with the relatively easy cases where the results follow to a large extent from straightforward application of the ift or simple calculation , and only then prove the harder theorems [ thmblockadd ] and [ thmintermediates ] .
@xmath165 theorem [ thmnewdepreac ] let the new reaction be @xmath390 so the new stoichiometric matrix is @xmath391 $ ] where @xmath392 .
define @xmath393 by @xmath394 .
we give the new reaction mass action kinetics and set its rate constant to be @xmath116 , so that the evolution of @xmath5 is governed by : @xmath395 the reduced jacobian matrix of @xmath74 w.r.t .
@xmath87 is @xmath396
so @xmath397 which is nonsingular as @xmath62 is a nondegenerate equilibrium of @xmath3 .
consequently , by corollary [ corift ] , there exists , for each sufficiently small @xmath398 , an equilibrium @xmath399 on @xmath400 and such that @xmath401 . by continuity of @xmath402 , @xmath399 is nondegenerate for sufficiently small @xmath398 , and linearly stable for @xmath5 if @xmath62 was linearly stable for @xmath3 .
the same arguments apply to give a nondegenerate equilibrium @xmath403 on @xmath400 and such that @xmath404 . as @xmath399 and @xmath399 are distinct for small @xmath405 , @xmath5 displays mpne , and mpse if @xmath62 and @xmath161 were linearly stable .
[ remnewdepreac ] it is clear from the proof that the added reaction in theorem [ thmnewdepreac ] can be given kinetics in any class provided the rate is a @xmath28 function of the concentrations and can be made arbitrarily small on any compact set in @xmath11 .
the theorem generalises to other structurally stable objects , notably hyperbolic periodic orbits .
theorem [ thmopenextension ] the stoichiometric matrix of @xmath5 can be taken to be @xmath406 $ ] where @xmath407 is the @xmath365 identity matrix .
let @xmath408 be any point on @xmath400 , and let the @xmath24th inflow - outflow reaction ( treated as a single reversible reaction ) have mass action kinetics with forward and backwards rate constants @xmath409 and @xmath116 respectively , so that @xmath410 is the rate of the @xmath24th added reaction .
evolution of @xmath5 is then governed by the odes @xmath411 .
let @xmath412 $ ] where the columns of @xmath413 are any basis for @xmath414 .
@xmath415 is nonsingular and hence its columns are a basis for @xmath9 .
define @xmath416 by @xmath417 .
define @xmath418 and @xmath419 by @xmath420 and @xmath421 .
note that @xmath422 . in this coordinate system equilibria of @xmath5 occur when @xmath423 ( as @xmath415 is nonsingular ) . for @xmath398
, @xmath74 has the same zeros as @xmath424 moreover @xmath425 ( as @xmath426 at @xmath427 , @xmath428 , and @xmath429 ) . differentiating gives
@xmath430 which is nonsingular as @xmath431 is . by the ift there exists , for sufficiently small @xmath398 , a @xmath28 curve @xmath432 of zeros of @xmath138 satisfying @xmath433 .
the same argument holds to give a @xmath28 curve @xmath434 of zeros of @xmath138 satisfying @xmath435 . by continuity of the determinant , they are nondegenerate zeros of @xmath138 for sufficiently small @xmath405 , and the same holds for @xmath74 as a quick calculation reveals that @xmath436 . since @xmath437 , @xmath438 and @xmath439 are positive and distinct for sufficiently small @xmath116 .
thus @xmath5 admits mpne .
the spectrum of @xmath440 is precisely @xmath441 with multiplicity @xmath442 with the remaining @xmath21 eigenvalues approaching the spectrum of @xmath443 as @xmath387 .
thus if @xmath62 is a linearly stable equilibrium of @xmath3 , then @xmath399 is a linearly stable equilibrium of @xmath5 for sufficiently small @xmath405 , and the same holds for @xmath161 . in other words ,
if @xmath3 admits mpse , then so does @xmath5 .
inflows and outflows were chosen to guarantee that @xmath400 remained invariant for @xmath5 and that for sufficiently small @xmath398 , the vector field of @xmath5 restricted to @xmath400 was @xmath116-close to that of @xmath3 restricted to @xmath400 in a natural sense .
nondegeneracy of @xmath62 and @xmath161 for @xmath3 ensured that equilibria @xmath399 and @xmath403 persisted on @xmath400 for sufficiently small @xmath444 and remained nondegenerate w.r.t .
moreover for @xmath398 the asymptotic stability of @xmath400 guaranteed that the equilibria @xmath399 and @xmath403 were also nondegenerate w.r.t .
directions transverse to @xmath400
. theorem [ thmtrivial ] let @xmath5 have reaction rates @xmath445 .
choose @xmath41 so that @xmath446 where @xmath447 is the stoichiometry of @xmath47 on the left of reaction @xmath25 ( assumed , w.l.o.g .
, to be irreversible ) .
we see that @xmath448 are the stoichiometric matrix of @xmath5 and a matrix whose columns are a basis for the stoichiometric subspace of @xmath5 respectively . @xmath449 and the evolution of @xmath5 is governed by @xmath450 since @xmath451 and @xmath452 , @xmath453 .
so @xmath454 is a positive equilibrium of @xmath5 , and likewise for @xmath455 .
moreover these equilibria are compatible as @xmath456 .
nondegeneracy is easy to confirm as @xmath457 recall that nondegeneracy of @xmath62 for @xmath3 is equivalent to nonsingularity of @xmath431 , and so @xmath454 is a nondegenerate equilibrium of @xmath5 .
a similar argument holds for @xmath455 , and so @xmath5 displays mpne . if @xmath62 is linearly stable w.r.t .
@xmath57 then , by ( [ eqtriv ] ) , @xmath454 is linearly stable w.r.t .
@xmath458 , and the same holds for @xmath161 and @xmath455 .
thus if @xmath3 admits mpse , then so does @xmath5 .
[ remkintriv ] in theorem [ thmtrivial ] , the vector field of @xmath5 restricted to @xmath459 is precisely that of @xmath3 , and so not just mpne , but all dynamical behaviours of @xmath3 are reproduced on this invariant set .
although this can be achieved naturally with mass action kinetics , it can also be achieved in other classes of kinetics where we can ensure that @xmath460 , for example any fixed power - law kinetics ( definition 4.8 in @xcite ) , or indeed various michaelis - menten or other hill - type kinetics . theorem [ thmnewwithopen ] let @xmath461 be the net stoichiometry change of @xmath47 in the @xmath24th reaction and @xmath462 .
give the new reaction @xmath242 mass action kinetics with both rate constants set to be @xmath463 .
the new rates @xmath41 for the other reactions are set as in theorem [ thmtrivial ] ( consistent with , but not assuming , mass action kinetics for @xmath3 ) , so that @xmath445 satisfies @xmath460 ( and @xmath464 ) .
define @xmath465 @xmath466 is the stoichiometric matrix of @xmath5 .
let @xmath467 .
@xmath5 gives rise to the following singularly perturbed system : @xmath468 for @xmath398 , @xmath469 has the same zeros as : @xmath470 moreover @xmath471 and we can calculate : @xmath472 using the fact that @xmath473 , we calculate @xmath474 . thus corollary [ corift ] gives , for sufficiently small @xmath398 , a @xmath28 curve @xmath475 of zeros of @xmath138 on @xmath476 , such that @xmath477 .
positivity of @xmath399 for sufficiently small @xmath116 is immediate as @xmath478 .
nondegeneracy of @xmath399 for sufficiently small @xmath398 follows from continuity of @xmath479 and the fact that @xmath480 . in a similar way we have , for sufficiently small @xmath398 , a curve @xmath403 of nondegenerate positive equilibria of @xmath5 on @xmath481 , and such that @xmath482 . as @xmath483 , @xmath399 and @xmath403
are compatible equilibria of @xmath5 .
thus @xmath5 admits mpne . if @xmath62 is linearly stable for @xmath3 , then @xmath399 is linearly stable for @xmath5 for sufficiently small @xmath398 .
define @xmath484 we can check quickly that if @xmath62 is linearly stable for @xmath3 , namely @xmath485 is hurwitz , then the hypotheses of lemma [ lemhurwitz ] are satisfied , and for sufficiently small @xmath444 all eigenvalues of @xmath363 lie in @xmath486 .
a similar argument applies to @xmath403 , and so if @xmath3 admits mpse , then so does @xmath5 .
[ remnewwithopen ] kinetics of the non - flow reactions in theorem [ thmnewwithopen ] can be chosen from various families other than mass action including any fixed power - law kinetics as described in remark [ remkintriv ] . indeed provided that fixing @xmath487 ( and perhaps choosing new constants ) gives back precisely the kinetics of @xmath3 , the working easily carries through .
the main conclusions of theorem [ thmnewwithopen ] survive if we replace a positive equilibrium which is nondegenerate w.r.t . its stoichiometric class by a positive periodic orbit which is hyperbolic w.r.t .
its stoichiometric class .
the generalisation is nontrivial , however , requiring results from geometric singular perturbation theory @xcite .
our next toy example illustrates why the proofs of theorems [ thmblockadd ] and [ thmintermediates ] can appear rather involved .
consider first a crn @xmath3 on a single species @xmath46 , with concentration @xmath73 and mass action rate constants shown : @xmath488 the evolution is governed by the differential equation @xmath489 and we immediately see that @xmath3 has equilibria at @xmath490 and @xmath491 .
the stoichiometric subspace is @xmath492 , and nondegeneracy simply means that the derivative of @xmath493 is nonvanishing at @xmath490 and @xmath491 , which is immediate .
so in fact @xmath3 admits mpne .
now consider adding in a new species @xmath47 ( with concentration @xmath201 ) and the reaction @xmath494 and setting forward and backward rate constants for the added reaction to be @xmath495 and @xmath43 to get the new crn @xmath5 : @xmath496 with dynamics : @xmath497 the stoichiometric subspace of @xmath5 is @xmath189 , and nondegeneracy of an equilibrium of @xmath5 just means that @xmath498 evaluated at the equilibrium is nonsingular .
intuition suggests that the modified crn might admit mpne : setting @xmath116 to @xmath33 gives two equilibria at @xmath499 and @xmath500 which we hope will persist for small @xmath398 .
however , @xmath501 and as both @xmath502 and @xmath503 are singular , the ift can not be used directly to construct families of nondegenerate positive equilibria @xmath504 as @xmath387 and @xmath505 as @xmath387 .
these equilibria exist , however : @xmath506 and @xmath507 are nondegenerate positive equilibria of @xmath5 for @xmath398 , and so @xmath5 admits mpne . the failure of a naive application of the ift in even such a simple example motivated the work in this paper .
although in this example the explicit calculations are easy enough , we illustrate a less explicit approach which gives insight into the proofs to follow .
the first step , similar to the argument in the proof of theorem [ thmnewwithopen ] , is to define a new family of vector fields @xmath508 tangent to stoichiometry classes of @xmath5 and where , for each fixed @xmath398 , @xmath508 has the same positive zeros as @xmath469 . to this end
define @xmath509 via : @xmath510 and note that @xmath511 and @xmath512 are interior points in the domain of @xmath138 satisfying @xmath513 .
the fact that @xmath138 is tangent to stoichiometry classes is in this case trivial as the stoichiometric subspace is @xmath189 .
we easily calculate that @xmath514 and @xmath515 are nonsingular , and so , by the ift , there exist , for sufficiently small @xmath398 , equilibria @xmath516 and @xmath517 s.t . @xmath518 and @xmath519 . by continuity , for sufficiently small @xmath405 , @xmath399 and @xmath403
are distinct , @xmath520 and @xmath521 , and so @xmath522 and @xmath523 .
so the curves consist of _ positive _ equilibria for @xmath398 sufficiently small .
we can check nondegeneracy of the equilibria directly by checking nonsingularity of @xmath524 and @xmath525 .
@xmath526 is easily seen to be nonzero for @xmath527 in neighbourhoods of @xmath43 and @xmath216 precisely by nondegeneracy of the original positive equilibria of @xmath3 at these points .
this shows that @xmath5 admits mpne .
this example illustrates some ideas in the proofs of theorems [ thmblockadd ] and [ thmintermediates ] to follow , but does not indicate the additional difficulties involved with adding several new reactions involving possibly more new species than new reactions , and where the stoichiometric subspace is not necessarily all of @xmath9 .
we begin with a lemma useful in both proofs : [ lemnondegen ] let @xmath20 be an @xmath22 matrix , @xmath87 be an @xmath528 matrix whose columns are a basis for @xmath57 , and @xmath88 a matrix defined by @xmath89 .
let @xmath240 be an @xmath323 matrix , and @xmath339 a @xmath529 matrix with rank @xmath249 .
let @xmath530 and @xmath531 be some @xmath28 functions .
define @xmath532 and @xmath533 by @xmath534 clearly @xmath535 .
suppose there is a @xmath28 curve @xmath536 satisfying @xmath537 for some @xmath538 , @xmath539 . define @xmath540 and @xmath541 .
assume that @xmath542 , and that @xmath543 is defined . 1 .
suppose that ( i ) @xmath544 , and ( ii ) @xmath545 is nonsingular
. then there exists @xmath546 s.t .
, for @xmath547 , @xmath548 .
2 . suppose that ( i ) @xmath549 is hurwitz stable , and ( ii ) @xmath545 is hurwitz stable
. then there exists @xmath546 s.t .
, for @xmath547 , @xmath550 is hurwitz stable .
define @xmath551 and note that @xmath415 has rank @xmath552 and that @xmath553 .
let @xmath554 .
we are interested in the reduced jacobian matrix : @xmath555 which is a representative of @xmath550 . 1 .
if @xmath556 is nonsingular then , as @xmath27 is @xmath28 , there exists @xmath557 s.t .
@xmath558 is nonsingular for @xmath559 .
then as @xmath387 , by lemma [ lemschur ] : @xmath560\,\mathrm{det}(m_\epsilon ) \to \mathrm{det}\,n_0\mathrm{det}\,m_0 \neq 0,\ ] ] as @xmath545 is nonsingular .
consequently , there exists @xmath561 s.t .
@xmath562 for @xmath563 .
2 . if @xmath545 and @xmath431 are hurwitz stable , then the assumptions of lemma [ lemhurwitz ] are satisfied , and @xmath564 is hurwitz stable .
this completes proof of the lemma .
we now prove theorems [ thmblockadd ] and [ thmintermediates ] .
geometrical insight into the proofs is given in appendix [ appgeom ] .
theorem [ thmblockadd ] let the @xmath249 added reversible reactions be @xmath565 define @xmath566 , with @xmath567 , @xmath568 and @xmath569 defined similarly . define @xmath392 and @xmath570 .
@xmath240 is now an @xmath323 matrix , and @xmath339 is a @xmath529 matrix with rank @xmath249 by assumption , implying @xmath250 .
, namely by reordering the species of @xmath47 if necessary , let the first @xmath249 rows of @xmath339 form a nonsingular square matrix so that @xmath571 , where @xmath572 is now a nonsingular @xmath366 matrix , and @xmath573 is a @xmath574 matrix .
we have @xmath575 where @xmath532 is the stoichiometric matrix of @xmath5 and @xmath415 is a basis for @xmath576 .
choose mass action kinetics for the added reactions and give the @xmath25th reaction forward and backward rate constants @xmath577 and @xmath578 respectively . setting @xmath579 and @xmath580 , the dynamics of the new system takes the form @xmath581 as @xmath339 has rank @xmath249 and hence represents an injective linear transformation , positive equilibria of @xmath5 are precisely the solutions of @xmath582 , namely of @xmath583 where @xmath584 .
write @xmath585 so that @xmath586 can be written @xmath587 . taking logs gives : @xmath588 we multiply through by @xmath589 which is defined by assumption , and let @xmath590 , @xmath591 . fix @xmath592 and @xmath593 where @xmath594 and @xmath595 refer to the top @xmath366 submatrices of @xmath568 and @xmath569 respectively . setting @xmath596 and exponentiating again
, we end up with : @xmath597 define @xmath598 by @xmath599 . with @xmath405 fixed and rate constants chosen as above , positive equilibria of @xmath5 occur precisely at @xmath600 note that @xmath601 and @xmath602 are in the domain of @xmath138 ( taken to be that of @xmath603 ) , which is open , and that @xmath604 .
we compute : @xmath605 now @xmath606 is the @xmath366 identity matrix and we easily confirm that @xmath607 and @xmath608 .
so , by the nonsingularity of @xmath431 and of @xmath572 , @xmath609 corollary [ corift ] gives , for sufficiently small @xmath405 , a @xmath28 curve @xmath610 of zeros of @xmath138 on @xmath611 such that @xmath612 .
for sufficiently small @xmath398 , @xmath613 ( as @xmath614 ) and @xmath615 ( as @xmath616 ) , and consequently @xmath617 , since @xmath618 thus each @xmath399 is a positive equilibrium of @xmath5 .
an identical argument replacing @xmath62 with @xmath161 gives a curve of positive equilibria @xmath403 on @xmath619 approaching @xmath620 as @xmath387 . since @xmath621 , @xmath622 and @xmath623
are compatible , and hence @xmath399 and @xmath403 are compatible .
we now infer nondegeneracy of @xmath399 using lemma [ lemnondegen](a ) .
define @xmath624 define @xmath625 by @xmath626 .
then , using ( [ eqyhat ] ) , we compute that @xmath627 .
observe that @xmath628 set @xmath629 , @xmath630 , @xmath631 ( which , by ( [ eqyhat ] ) , is defined and nonnegative ) , and @xmath632 .
note that @xmath633 .
we compute that , as required in lemma [ lemnondegen ] , @xmath634 since @xmath635 as @xmath387 and @xmath636 as @xmath637 . by ( [ eqyhat ] ) , @xmath638 = p^{-\mathbf{1}^\mathrm{t}\gamma } > 0 $ ] , while @xmath639 = 0 $ ] for @xmath640 , @xmath641 , @xmath642 .
applying the cauchy - binet formula to @xmath545 gives , @xmath643\beta[\theta|\langle m \rangle]^2 = ( -1)^mc_0k'\beta[\langle m \rangle]^2\neq 0,\ ] ] and so @xmath545 is nonsingular .
all the conditions of lemma [ lemnondegen](a ) are met , and we conclude that for each sufficiently small @xmath405 , @xmath399 is a nondegenerate equilibrium of @xmath5 . an identical argument applies to @xmath403 .
thus @xmath5 admits mpne . to see
inheritance of multistability observe that @xmath644 @xmath545 is hence similar to the negative semidefinite matrix @xmath645 .
as @xmath545 has already been shown to be nonsingular , it must in fact be hurwitz . if @xmath62 is linearly stable for @xmath3 , namely @xmath549 is hurwitz stable , then the conditions of lemma [ lemnondegen](b ) are met , and we conclude that for each sufficiently small @xmath116 , @xmath399 is linearly stable for @xmath5 .
an identical argument applies to @xmath403 , and so if @xmath3 admits mpse , then so does @xmath5 .
theorem [ thmintermediates ] the proof is similar to that of theorem [ thmblockadd ] .
recall that we create @xmath5 from @xmath3 by replacing each of the @xmath249 reactions : @xmath646 as we are assuming that @xmath647 has rank @xmath249 , w.l.o.g .
( namely , by reordering the species of @xmath47 if necessary ) , we write @xmath571 where @xmath572 is a nonsingular @xmath366 matrix , and @xmath573 is a @xmath574 matrix . also w.l.o.g ( namely ,
by splitting and/or reordering the reactions of @xmath3 as necessary ) let the reaction @xmath648 figure as the @xmath24th reaction in the original crn , proceeding with rate @xmath649 , and let @xmath650 .
set the rate of each new reaction @xmath651 to be @xmath649 ( this is consistent with mass action kinetics if the original reaction had mass action kinetics ) and set the second added reaction in the @xmath24th chain to have mass action kinetics with rate constant @xmath652 ( @xmath268 ) .
define @xmath653 and @xmath654 .
note that for @xmath14 , @xmath655 under the assumption of weak general kinetics , and in particular @xmath656 .
define @xmath657 , @xmath658 , @xmath659 , @xmath660 so that @xmath553 .
@xmath532 is the stoichiometric matrix of @xmath5 and the columns of @xmath415 are a basis for @xmath576 .
the dynamics of @xmath5 is governed by @xmath661 thus , with the kinetic choices made so far , the net effect on the vector field of `` inserting '' the complexes @xmath662 ( @xmath663 ) is the same adding to @xmath3 @xmath249 new ( pseudo)-reactions with stoichiometric matrix @xmath664 and rate vector @xmath665 .
as @xmath339 has rank @xmath249 ( and hence corresponds to an injective linear transformation ) , positive equilibria of @xmath5 are solutions of @xmath666 write @xmath585 .
the condition @xmath667 reads @xmath668 .
we now perform some manipulations similar to those in the proof of theorem [ thmblockadd ] . setting @xmath669 , and defining @xmath670 , @xmath671 , and @xmath672
, we end up with @xmath673 define @xmath598 by @xmath674 .
then , for @xmath398 and reaction rates chosen as above , positive equilibria of @xmath5 occur precisely at @xmath600 note that @xmath601 and @xmath602 are in the domain of @xmath138 ( taken to be that of @xmath603 ) , which is open , and that @xmath604 .
we compute : @xmath675 now @xmath676 , @xmath607 , @xmath608 , and so , as @xmath572 and @xmath431 are nonsingular , @xmath677 .
corollary [ corift ] gives , for sufficiently small @xmath405 , a curve @xmath610 of zeros of @xmath138 on @xmath611 such that @xmath612 .
for sufficiently small @xmath405 , @xmath613 , @xmath678 ( as @xmath679 ) and @xmath615 ( as @xmath680 ) , and consequently @xmath617 as @xmath681 thus the curve @xmath399 consists of positive equilibria of @xmath5 . an identical argument replacing @xmath62 with @xmath161
gives positive equilibria @xmath403 on @xmath619 approaching @xmath620 as @xmath387 . since @xmath621 , @xmath399 and @xmath403
are compatible .
we can now infer nondegeneracy of @xmath399 and @xmath403 using lemma [ lemnondegen](a ) .
set @xmath682 define @xmath683\,,\\ b_\epsilon:=\epsilon d_yf(x(\epsilon ) , y(\epsilon ) ; \epsilon ) = -\mathrm{diag}(\underline{v}(x(\epsilon)))\beta^\mathrm{t}\mathrm{diag}(\bm{\epsilon}/y(\epsilon))\ , .
\end{array}\ ] ] set @xmath684 , @xmath630 and @xmath631 ( which , by ( [ eqyhat1 ] ) , is defined and nonnegative ) .
note that @xmath633 .
it is clear that @xmath685 as @xmath387 as the quantity in the square bracket approaches the constant matrix @xmath686 $ ] as @xmath387 . from ( [ eqyhat1 ] ) , @xmath687 $ ] is a positive constant , while @xmath639 = 0 $ ] for @xmath640 , @xmath641 , @xmath642 .
then @xmath688 , and by the cauchy - binet formula , @xmath689(\beta[\theta|\langle m \rangle])^2 \to ( -1)^mc_0k'\beta[\langle m \rangle]^2\neq 0,\ ] ] and so @xmath545 is nonsingular .
all the conditions of lemma [ lemnondegen](a ) are met , and we conclude that for each sufficiently small @xmath405 , @xmath399 is a nondegenerate equilibrium of @xmath5 . an identical argument applies to @xmath403 .
almost identically to the calculation in the proof of theorem [ thmblockadd ] , @xmath545 can be seen to be hurwitz stable . if @xmath62 is linearly stable @xmath3 , namely @xmath549 is hurwitz stable , then by lemma [ lemnondegen](b ) , for each sufficiently small @xmath405 , @xmath399 is linearly stable with respect to @xmath458 .
an identical argument applies to @xmath403 , and so if @xmath3 admits mpse , then so does @xmath5 .
[ remkin ] examining the proofs of theorems [ thmblockadd ] and [ thmintermediates ] we see that mass action kinetics does not really play a fundamental role .
most important are the equations ( [ eqyhat ] ) and ( [ eqyhat1 ] ) , namely that we can sufficiently control the rates of the added reactions to ensure that the @xmath690 values at the new equilibria approach @xmath33 as some parameter @xmath387 .
a variety of kinetic assumptions would allow this control , including other power - law kinetics ( definition 4.8 in @xcite ) .
the results admit generalisation in this direction noting , however , that the full - rank conditions in the theorems would now apply both to the blocks added to the stoichiometry matrices , and to the matrices of exponents .
we have begun to describe how a crn may be enlarged while maintaining the properties of having multiple positive nondegenerate equilibria ( mpne ) , or multiple positive linearly stable equilibria ( mpse ) .
the modifications in theorems [ thmnewdepreac ] to [ thmintermediates ] collectively define a partial order @xmath6 on the set of all crns . in the biologically motivated example of section [ secbioexample ] we identify an `` atom of mpse '' , namely a crn which admits mpse and which is minimal w.r.t .
@xmath6 , and show how this atom occurs in various published models , immediately implying mpse in these models .
we suspect that often the small multistationary networks described in @xcite , minimal w.r.t .
the induced subnetwork partial order , are also minimal w.r.t .
@xmath6 , and hence form natural building - blocks of crns displaying mpne .
this remains to be confirmed .
our approach was local : the main theoretical tool used was the ift and , informally speaking , we mostly constructed @xmath5 as a _ perturbation _ of @xmath3 in some sense .
the results presented in this paper certainly do not exhaust the possibilities in this approach .
we have chosen not to include a number of partial or isolated results in the same vein as theorems [ thmblockadd ] and [ thmintermediates ] .
these theorems themselves can probably be strengthened ; for example , with some added effort , reversibility of the added reactions in theorem [ thmblockadd ] can probably be replaced with weak reversibility of the added reactions .
it also seems that insisting that the original crn should have mass action kinetics may allow modifications which our basic assumptions would not permit because it forces certain relationships to hold between reaction rates ( for example , the ratio of the rates of two reactions with the same source complex would be constant ) .
further , as remarked on at several points , the results permit modification in various directions where we choose classes of kinetics other than mass action .
going beyond the techniques here , it seems likely that algebraic approaches may reveal network modifications which preserve mpe or mpne , but which could not be predicted using a local approach : some of the modifications in section [ secextended ] and the results in @xcite point in this direction .
ultimately the existence of mpe for mass action systems is about the cardinality of the intersection of certain algebraic varieties : it may be that adding new species or reactions in certain ways can be proved to allow mpe , but not by a local approach . working towards the `` best '' partial order on crns from the point of view of the inheritance of mpe or mpne
is a project to be continued .
the evolution of @xmath691 $ ] , @xmath692 $ ] , @xmath693 $ ] , @xmath694 $ ] and @xmath695 $ ] is governed by the ode system : @xmath696 it is easy to check that the system has precisely two independent conserved quantities @xmath697 where @xmath69 and @xmath698 are positive reals . at steady state
, one can easily express all variables in terms of @xmath699 : @xmath700 note that positive values of @xmath699 will automatically give rise to positive values for all variables except @xmath701 . substituting the expressions above into the expression for @xmath702 and
clearing denominators yields a cubic equation in @xmath699 : @xmath703x_2 ^ 2 \\
\nonumber & & + [ 2n { k_1 } { k_3 } { k_5 } { k_6}- { k_3}^2 { k_5 } { k_6}-2 { k_2 } { k_3 } { k_5 } { k_6}-m { k_1 } { k_3 } { k_5 } { k_6}- { k_2}^2 { k_5 } { k_6}\\ \nonumber & & + 2n { k_1 } { k_2 } { k_5 } { k_6 } -m { k_1 } { k_2 } { k_5 } { k_6}-m { k_1 } { k_3}^2 { k_5}-m { k_1 } { k_2 } { k_3 } { k_5}-m^2 { k_1 } { k_3}^2 { k_4}\\ \nonumber & & -m^2 { k_1 } { k_2 } { k_3 } { k_4 } ] x_2 + k_5k_6(k_2+k_3)^2n = 0\end{aligned}\ ] ] now we impose that the coefficients of the cubic above have alternate signs ( a necessary condition for existence of three positive roots ) , and by trial and error we identify values for parameters that indeed give rise to distinct positive roots , and which moreover make @xmath704 in the steady state expression for @xmath701 . for example , @xmath705 gives @xmath706 , and it can be checked that @xmath707 , so that @xmath708 has a positive root in each one of the intervals @xmath709 . from the steady state expression for @xmath701 we get @xmath710 where @xmath711 . since @xmath712 and @xmath713 it follows that @xmath714 on @xmath715 , and therefore @xmath714 for all roots @xmath73 of @xmath708 .
therefore all three @xmath699 values lead to positive @xmath701 values . finally , for this choice of parameters the reduced determinant of the system calculated in terms of @xmath699 is equal to @xmath716 , which ensures that all three steady states are nondegenerate .
with the constants chosen as above , computing the steady states numerically yields : @xmath717 it was checked numerically that the first and the third steady state are linearly stable w.r.t . their stoichiometry class ( whereas the second is not ) and thus this crn admits mpse .
we present a geometrical interpretation of theorems [ thmblockadd ] and [ thmintermediates ] , which we hope can in future be used to use to generate further theorems in these directions . for simplicity
we treat only the case of precisely @xmath249 new species entering nondegenerately into @xmath249 new reactions ( theorem [ thmblockadd ] ) or @xmath249 new chains ( theorem [ thmintermediates ] ) , although the ideas extend , with some additional effort , to the case where the number of new species is greater than the number of new reactions .
[ deftransverse ] let @xmath718 be open .
a @xmath28 map @xmath719 is _ transverse _ to a smooth manifold @xmath720 if @xmath721 implies that @xmath722 .
we write @xmath723 .
if @xmath724 , we say that @xmath708 is transverse to @xmath69 along @xmath725 , denoted @xmath726 , if @xmath727 implies that @xmath722 .
[ lemtransverse ] let @xmath728 be open and bounded , @xmath729 a @xmath28 immersion , and @xmath730 some coset of a linear subspace . @xmath731 and @xmath732 are some norms on @xmath9 and on linear maps from @xmath733 respectively
let @xmath734 be any closed set with nonempty interior @xmath735 , and assume that @xmath736 .
given @xmath398 , define @xmath737 ( @xmath738 is a neighbourhood of @xmath708 in the @xmath28 compact - open topology , which we abbreviate as the @xmath28 topology . ) assume that @xmath739 consists of @xmath740 distinct points and @xmath741 .
then there exists @xmath143 s.t .
@xmath742 implies that : ( i ) @xmath743 ; and ( ii ) @xmath744 consists of @xmath745 distinct points .
note that @xmath725 is compact , and @xmath67 is closed .
( i ) is now a consequence of theorem 2.1 in chapter 3 of @xcite which implies that the set of @xmath28 functions from @xmath746 transverse to @xmath67 along @xmath725 is open ( and dense ) in the @xmath28 topology .
\(ii ) by ( i ) choose @xmath747 s.t .
@xmath748 implies that @xmath743 . by a variant of the regular value theorem ( thm .
3.3 in chapter 1 of @xcite for example ) , the facts that @xmath749 is @xmath16-dimensional in @xmath750 , that @xmath736 , and that @xmath751 is a zero dimensional manifold ( a nonempty set of isolated points ) together imply that @xmath752 .
this in turn implies , using ( i ) and the same variant of the regular value theorem , that @xmath753 is a zero dimensional manifold or possibly empty for each @xmath748 .
next , let @xmath754 be a linear embedding with image @xmath67 , and define @xmath755 by @xmath756 .
by construction , zeros of @xmath138 are in one to one correspondence with points in @xmath757 .
as @xmath757 does not intersect @xmath758 and @xmath758 is compact and @xmath67 closed @xmath759 so , for @xmath760 , @xmath761 and so @xmath762 . let @xmath763 .
consider any path @xmath764 \ni t \mapsto f_t \in n_{\epsilon_0,l}(f)$ ] with @xmath765 .
define @xmath766\to \mathbb{r}^{n+m}$ ] by @xmath767 .
@xmath768 is compact : that it is closed follows from continuity of @xmath769 ; as @xmath770 is continuous in both @xmath771 and @xmath42 , @xmath772 \times l\}$ ] is a bounded subset of @xmath750 and as @xmath773 is defined and continuous on @xmath67 , @xmath774 is also bounded . for each fixed @xmath771
, @xmath775 has zeros in one - to - one correspondence with the points of @xmath776 .
define @xmath777 .
we show that @xmath778 is locally constant , and hence constant on [ 0,1 ] .
suppose @xmath779 for some @xmath780 $ ] .
\(a ) @xmath778 is locally nondecreasing .
the transversality condition at some @xmath781 , namely @xmath782 , is equivalent to nonsingularity of @xmath783 .
so , by the ift , there exists @xmath784 and an open neighbourhood @xmath785 of the zero set of @xmath786 such that @xmath787 $ ] implies that @xmath788 contains exactly @xmath21 zeros in @xmath789 .
consequently , @xmath790 for @xmath787 $ ] .
\(b ) @xmath778 is locally nonincreasing .
suppose there exists a sequence @xmath791 $ ] with @xmath792 and s.t .
@xmath793 for each @xmath24 . by the argument in ( a ) for each @xmath794 there exists @xmath795 s.t .
as @xmath797 is compact , by passing to a subsequence if necessary , we may assume that @xmath798 .
as @xmath799 ( by construction of @xmath800 ) , we must have @xmath801 . by continuity of @xmath769 , @xmath802 , contradicting @xmath779 .
+ + as @xmath803 and @xmath804 is locally constant on @xmath764 $ ] , @xmath805 for all @xmath806 $ ] .
since the path @xmath807 was arbitrary , @xmath808 for all @xmath742 .
we return to the geometry of the proofs of theorems [ thmblockadd ] and [ thmintermediates ] focussing attention on @xmath62 , a nondegenerate equilibrium of @xmath3 .
most notation is as in section [ secthms ] .
we write @xmath809 for the concentrations of species in @xmath5 .
let @xmath810 be the natural embedding of @xmath9 ( the state space of @xmath3 ) in @xmath750 ( the state space of @xmath5 ) and denote the images of objects under @xmath810 with an overline .
so , for example , @xmath811 .
let @xmath264 be the positive equilibrium set of @xmath3 , namely @xmath812 .
let @xmath236 , @xmath576 and @xmath813 be the span of the new reaction vectors , namely the image of @xmath664 in theorems [ thmblockadd ] and [ thmintermediates ] .
recall that in both of theorems [ thmblockadd ] and [ thmintermediates ] , we found that positive equilibria of @xmath5 were exactly those points satisfying @xmath814 and an equation of the form @xmath815 ( equations [ eqyhat ] and [ eqyhat1 ] ) .
since here we are assuming for simplicity that @xmath816 is empty , the latter equation becomes @xmath817 for some @xmath28 function @xmath818 .
thus for each fixed @xmath405 , the positive equilibrium set of @xmath5 is @xmath819 observe that @xmath820 is the graph of @xmath821 , and that , by continuity of @xmath822 , @xmath823 converges in the @xmath28 topology to the zero function as @xmath387 .
let @xmath824 , @xmath825 , @xmath826 , and @xmath827 1 .
* claim 1 . *
_ there is an open neighbourhood @xmath828 of @xmath62 s.t .
@xmath829 is an @xmath442 dimensional smooth submanifold of @xmath9 intersecting @xmath400 only at @xmath62 .
moreover @xmath830 is the image of a map transverse to @xmath400 . _
these statements are just the geometrical interpretation of nondegeneracy of @xmath62 using the ift . to see this , write vectors of @xmath9 in the form @xmath831 where @xmath832 and @xmath833 ( any other complementary subspace will do ) .
if @xmath834 is the vector field of @xmath3 , nondegeneracy of @xmath62 is equivalent to nonsingularity of @xmath835 .
writing @xmath836 , the ift tells us that there exist neighbourhoods @xmath837 of @xmath838 in @xmath839 and @xmath141 of @xmath840 in @xmath841 , and a @xmath28 function @xmath842 satisfying @xmath843 , such that for @xmath844 , @xmath845 if and only if @xmath846 .
@xmath847 is then a portion of @xmath264 containing @xmath62 , and lying entirely in @xmath11 .
as it is precisely the graph of @xmath111 , it intersects @xmath400 only at @xmath62 .
define @xmath848 , @xmath849 , so that @xmath850 is a @xmath28 injection with image @xmath830 .
@xmath851 , and @xmath852 ( as @xmath850 is an injection ) .
since @xmath853 includes no nontrivial element of @xmath854 , @xmath855 namely @xmath856 . the situation in @xmath857 coordinates is illustrated schematically in figure [ figtrans](a ) .
* claim 2 . *
_ @xmath858 is an @xmath859-dimensional submanifold of @xmath750 intersecting @xmath860 only at @xmath861 .
moreover @xmath858 is the image of a map transverse to @xmath860 at @xmath861 .
_ the only thing that needs checking is transversality : the remaining claims are immediate from claim 1 as @xmath862 , and @xmath863 ( as @xmath864 is trivial ) .
define the @xmath28 injection @xmath865 by @xmath866 , namely @xmath867 .
@xmath868 has image @xmath858 .
note that @xmath869 and @xmath870 . by the nondegeneracy assumptions on the new reactions , @xmath871 , and @xmath872 .
moreover , by embedding of the objects of claim 1 in @xmath750 , @xmath873 .
thus @xmath874 proving that @xmath875 .
claim 3 . *
_ for small @xmath398 and near @xmath861 , @xmath820 is an @xmath442 dimensional submanifold of @xmath750 which is the image of a map @xmath876 , @xmath28 close to @xmath868 . _
define @xmath877 by @xmath878 .
@xmath876 is a @xmath28 injection as @xmath868 is a @xmath28 injection and @xmath822 is @xmath28 .
define @xmath879 by construction @xmath880 lies in @xmath881 and as a @xmath28 embedding of @xmath837 in @xmath750 , it is an @xmath442 dimensional @xmath28 submanifold of @xmath750 .
moreover @xmath882 which converges to the zero function in the @xmath28 topology as @xmath883 .
* claim 4 . *
_ let @xmath725 be any closed neighbourhood of @xmath838 in @xmath837 and let @xmath884 .
then for any sufficiently small @xmath405 , @xmath885 , and there exists a positive @xmath399 such that @xmath886 .
@xmath399 is isolated , namely there is a neighbourhood of @xmath399 in @xmath860 which contains no other elements of @xmath820 .
_ @xmath885 follows from @xmath887 and lemma [ lemtransverse](i ) . that @xmath888 is nonempty and consists of a single point follows from lemma [ lemtransverse](ii ) . as @xmath820 ( and hence @xmath889 ) lies in @xmath881
, @xmath399 is positive . for ( any ) fixed @xmath890 ,
consider the open neighbourhood @xmath891 by choosing @xmath116 small enough we can ensure that @xmath892 for @xmath893 , and so @xmath894 . then @xmath895 is a neighbourhood of @xmath399 in @xmath860 , and @xmath399 is the unique element of @xmath820 in @xmath895 , confirming that @xmath399 is isolated .
the construction of @xmath399 is illustrated schematically in figure [ figtrans](b ) .
( -0.2,0 ) ( 5.8,0 ) ; ( 1.5,0 ) ( 4.5,0 ) ; ( 1.5,0 ) ( 4.5,0 ) ; at ( 2.2,0.2 ) @xmath837 ; ( 0,-0.2 ) ( 0,4 ) ; ( 0,1 ) ( 0,3 ) ; ( 0,1 ) ( 0,3 ) ; at ( 0.2,1.5 ) @xmath141 ; at ( 6,-0.2 ) @xmath194 ; at ( -0.2,4.2 ) @xmath41 ; at ( 3.3,4 ) @xmath400 ; ( -0.2,0 ) ( 6,0 ) ; ( 1.5,0 ) ( 4.5,0 ) ; ( 1.5,0 ) ( 4.5,0 ) ; at ( 2.4,0.2 ) @xmath900 ; at ( 1.8,1.4 ) @xmath901 ; ( -0.2,-0.2 )
( 4,4 ) ; ( 1,1 )
( 3,3 ) ; ( 1,1 )
( 3,3 ) ; ( 0,-0.2 ) ( 0,5 ) ; at ( 6,-0.2 ) @xmath194 ; at ( 4,4.2 ) @xmath41 ; at ( -0.2,5 ) @xmath201 ; at ( 6.5,3.3 ) @xmath902 ; ( 5,2 ) ( 5,2.35 ) ; ( 6.8,2.3 )
( 6.8,2.6 ) ; ( 5.5,1 ) ( 5.5,2.8 ) ; ( 5.5,2.8 )
( 7.5,4.8 ) ; ( 3.5,2.8 )
( 5.5,2.8 ) ; ( 7.7,3 )
( 7.7,3.6 ) ; ( 7.7,4.2 )
( 7.7,4.8 ) ; at ( 7.7,3.9 ) @xmath771 ; ( 4,1 ) ( 3.55,2.8 )
( 5.55,4.8 )
( 6,3 ) cycle ; a similar argument applies to @xmath161 : we get , for sufficiently small @xmath398 , the existence of a positive equilibrium @xmath403 near to @xmath906 on @xmath907 . for sufficiently small @xmath116 ,
@xmath399 and @xmath403 are distinct , and as @xmath908 , they are compatible .
the above arguments do not automatically give us nondegeneracy of @xmath399 and @xmath403 , but we do get that these are isolated equilibria ( on @xmath909 ) and so nondegeneracy should not be a surprise . | we study how the properties of allowing multiple positive nondegenerate equilibria ( mpne ) and multiple positive linearly stable equilibria ( mpse ) are inherited in chemical reaction networks ( crns ) . specifically , when is it that we can deduce that a crn admits mpne or mpse based on analysis of its subnetworks ? using basic techniques from analysis we are able to identify a number of situations where mpne and mpse
are `` inherited '' as we build up a network .
some of these modifications are known while others are new , but all results are proved using the same basic framework , which we believe will yield further results .
the results are presented primarily for mass action kinetics , although with natural , and in some cases immediate , generalisation to other classes of kinetics .
multiple equilibria ; chemical reaction networks * msc . * 80a30 ; 15a15 ; 37c25 |
star formation occurs across the galactic disc with most molecular material concentrated at low latitudes @xmath1 .
both in the inner and in the outer galaxy , young stars still partly embedded in the dense gas and dust in molecular clouds have been found ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
they represent current active star formation sites .
the study and census of star formation sites in regions of higher galactic latitudes have received comparatively little attention and coverage .
however , the detection and characterisation of star formation sites across the whole galaxy have strong implications on its structure and evolution . in our previous studies of star formation in the outer galaxy ,
we have reported discovery of what can be designated as star formation sites in extreme environments , namely near the distant `` edge '' of the galactic disc ( at galactocentric distances of about 16.5 kpc @xcite ) , or far below the midplane of the galactic disc ( at vertical distances of about 500 pc @xcite ) . in both cases ,
the presence of molecular material and of clusters of embedded stars in these environments posit that the physical conditions required to form a cluster of stars are met in locations at distances away from the galactic plane ( or away from the galactic centre ) larger than reasonably expected .
iras 06345 - 3023 is an iras psc source in the outer galaxy that currently appears classified in the simbad data base as `` 6dfgs gj063630.0 - 302542 galaxy '' .
this means that it is listed in the 6dfgs , the six - degree field galaxy survey catalogue as being an extragalactic source .
however , a search in the 6dfgs catalogue , at the iras coordinates , does not return this source .
instead , a different source ( g0636403 - 302842 ) is presented , which is located about 4 arcminutes away . in addition , ned ( the nasa extragalactic database ) does not contain this source .
@xcite first called attention on the presence of two nebulous `` stars '' on eso / src j plates at this location .
@xcite used the 4 metre nanten milimetre telescope to study a sample of intermediate - to - high latitude iras sources and reported detection of co emission from iras 06345 - 3023 .
their co maps , although at a coarse angular resolution , show a clear peak at the position of the source . as part of our study of young
embedded clusters in the outer galaxy ( e.g. * ? ? ?
* ; * ? ? ?
* ) , we have conducted observations ( near - infrared @xmath0 imaging , and millimetre co line ) towards iras 06345 - 3023 .
these observations revealed the presence of young stars embedded in a molecular cloud core .
we report here our near - infrared discovery of an aggregate of young stars seen towards iras 06345 - 3023 .
in addition , we use new co data , to characterise the molecular environment and derive its kinematic distance .
section 2 describes the observations and data reduction . in section 3
, we present and discuss the results .
a summary is given in section 4 .
near - infrared ( @xmath2 , @xmath3 and @xmath4 ) images were obtained on 2000 november 12 and on 2002 january 8 using the eso antu ( vlt unit 1 ) telescope equipped with the short - wavelength arm ( hawaii rockwell ) of the isaac instrument .
the isaac camera @xcite contains a 1024 @xmath5 1024 pixel near - infrared array and was used with a plate scale of 0.147 arcsec / pixel resulting in a field of view of 2.5 @xmath5 2.5 arcmin@xmath6 on the sky . for each filter , dithered sky positions were observed .
a series of 12 images with individual on - source integration time of 6 and 3.55 seconds was taken in the @xmath2 and in the @xmath3 bands , respectively . similarly ,
series of 6 images , each of 2 second integration time , were obtained in the @xmath4 band .
the images were reduced with the image reduction and analysis facility ( iraf ) , using a set of our own scripts to correct for bad pixels , subtract the sky background , and flatfield the images .
dome flats were used to correct for the pixel - to - pixel variations of the response .
the selected images were then aligned , shifted , trimmed , and co - added to produce a final mosaic image for each band @xmath0 .
correction for bad pixels was made while constructing the final mosaics that cover about @xmath7 arcmin@xmath6 on the sky .
point sources were extracted using daofind with a detection threshold of 5@xmath8 .
the images were inspected to look for false detections that had been included by daofind in the list of detected sources .
these sources were eliminated from the source list .
aperture photometry was made with a small aperture ( radius = 0.5@xmath9 ) and aperture corrections , found from bright and isolated stars in each image , were used to correct for the flux lost in the wings of the psf .
the error in determining the aperture correction was @xmath10 0.03 mag in all cases .
we used the 2mass all - sky release point source catalogue @xcite to calibrate our observations .
the @xmath0 zeropoints were determined using 2mass stars brighter than @xmath4 = 14.7 mag .
the standard deviations of the offsets between isaac and 2mass photometry are 0.1 mag , in all bands .
given the relatively small number of stars in these images , it is not possible to make a robust estimate of the completeness limit of the observations based on the statistics of the stars detected .
however , a relatively good determination of the completeness limit can be obtained calculating _
e.g. _ the @xmath11-limit in each band . those are 20.6 in @xmath2 , 20.0 in @xmath3 , and 18.9 mag in @xmath4 . as part of a survey of molecular gas in the third galactic quadrant
, the region around the position of the iras source was mapped using the single - dish 15-m sest radiotelescope in 1999 september .
the map was obtained in the rotational line of @xmath12co(1 - 0 ) ( 115.271 ghz ) and consisted of @xmath13 positions , in full - beam spacing ( @xmath14 ) , and centred on the iras coordinates .
the integration time was 60 seconds .
the spectra were taken in frequency switching mode , recommended to save observational time when mapping extended sources .
one additional central spectrum was taken in beam - switching mode , with integration time of 120 seconds , in order to obtain improved baselines to search for the presence of molecular outflows .
a high - resolution 2000 channel acousto - optical spectrometer was used as a back end , with a total bandwidth of 86 mhz and a channel width of 43 khz , which at the frequency of 115 ghz corresponds to approximately 0.11 km s@xmath15 .
the antenna temperature was calibrated with the standard chopper wheel method .
pointing was checked regularly towards known circumstellar sio masers , and pointing accuracy was estimated to be better than @xmath16 .
the data reduction pipeline was composed of the following steps : _
i ) _ folding the frequency - switched spectrum ; _ ii ) _ fitting the baseline by a polynomial and subtracting it ; _ iii ) _ obtaining the main - beam temperature @xmath17 by dividing the antenna temperature @xmath18 by the @xmath19 factor , equal to 0.7 . the spectrum baseline rms noise ( in @xmath17 ) ,
averaged over all map positions , has been found to be 0.15 k. ( blue ) , @xmath3 ( green ) , and @xmath4 ( red ) colour composite image towards iras 06345 - 3023 covering @xmath20 .
north is up and east to the left .
notice the nebular emission around a small concentration of red stars towards the centre of the image , and the very small number of field stars.,width=291 ]
figure [ rgb995 ] presents the vlt / isaac @xmath0 near - infrared colour composite image obtained towards iras 06345 - 3023 .
centred on the image , extended nebular emission involving a few bright stars is displayed .
in addition , when compared to other fields , usually at low galactic latitudes , we notice a very small number of field stars , either background or foreground , that populate the galactic disc and frequently contaminate deep near - infrared images
. a close look at the image ( see the zoomed - in view of the same region shown in figure [ rgb995 - 1 ] ) reveals the presence of a butterfly - shaped bipolar dark cloud , to the north - northeast and to the south - southwest .
notice also the bright rims of the dark `` butterfly wings '' illuminated by the central stars .
the richness of the structures seen includes a quasi - circular bright ring and bright wisps or rays of light against the black opaque northern wing ( see also fig .
[ rgb999 ] for the nomenclature used here ) . in fig .
[ rgb999 ] we reproduced the colour composite image using a colour coding emphasizing the bright stars as opposed to the diffuse nebular emission .
a few important points can be derived as follows .
the bright source labeled `` wispy neb '' is not a point source , it corresponds to pure diffuse emission .
several wisps in this nebula ( better distiguished in fig .
[ rgb995 - 1 ] ) seem to point away from a common point , the location of one of the reddest sources , that labeled `` irs 4 '' .
the bluer colour of the wispy nebula supports the idea that it corresponds to scattered light from a cavity located close to the centre of the butterfly - shaped core where opacity is lower . in fact , the position of one of the two `` nebulous stars '' that were noticed on the optical eso / src images @xcite coincides with the position of this wispy nebula .
the other optical `` nebulous star '' corresponds to the blue emission seen to the west including the circular arc ( the `` arc '' ) .
this arc - like structure appears to be associated with the source labeled `` irs 3 '' which is the brightest red embedded source present in the images .
the arc is quasi - circular , very well delineated , and constitutes an engimatic structure , similar to the structure seen around a few other young stellar objects .
examples of these are the diamond - ring object of @xcite , which was conjectured to represent a circumbinary structure existing during a brief stage in the formation of a binary star ; and gg tau , a multiple young source , which exhibits a circumbinary ring ( e.g. * ? ? ?
* ; * ? ? ?
* ) , and that has recently been suggested to be forming planets @xcite .
irs 3 could be another case of a multiple young object denouncing its multiplicity through the presence of circumbinary arcs or rings , seen in the near - infrared or submillimetre .
interestingly , neither of the two optical `` nebulous stars '' is a real star ( point source ) but instead corresponds to diffuse scattered emission escaping through lines - of - sight of lower optical depth .
several cavities excavated by winds from young stars are now known and have been modelled ( e.g. * ? ? ?
* ; * ? ? ?
our images seem to confirm the rule that young stars disrupt the parent cloud core reducing the opacity from specific lines - of - sight and resulting in very non - homogenous distributions of extinction across cloud cores . also remarkable
is the detection of wisps of bluer light escaping from this star formation site , similar to sunlight under low - lying clouds on a partly cloudy day on earth .
in addition , despite its faintness , a careful look reveals the presence of a bright rim of this molecular cloud core specially in the southwest part of the dark nebula ( fig .
[ rgb995 - 1 ] ) . but with a colour coding emphasizing the emission from bright stars .
notice the nomenclature of the structures labeled here , some of which can be seen more clearly in fig.[rgb995 - 1],width=302 ] we have detected co emission towards the positions of the nebular emission and of this group of stars .
moreover , the location of the nebula coincides with the position where the molecular gas peaks .
[ co1 ] shows the beam switching spectrum from the iras 063453023 revealing a clear detection of a co line . a gaussian fit yields a @xmath21 peak value of 8.1 k at the @xmath22 km s@xmath15 in good agreement with @xcite .
the line presents some deviation from a gaussian , exhibiting moderate wings that could be due to a molecular outflow .
our @xmath23 co map ( fig . [ comap ] ) shows relatively constant co emission , but peaking at the centre position where the iras source is located .
because the map was obtained in frequency - switching mode , we did not try to look for the presence of outflows in the beams of this map .
either adopting a flat rotation curve in the outer galaxy or using a circular rotation model , a heliocentric distance @xmath24 kpc , and a galactocentric distance @xmath25 kpc are found , respectively .
this is in good agreement with the distance quoted by @xcite .
thus , the projected size of the region occupied by the near - infrared nebula and butterfly - wings turns out to be @xmath26 pc , a typical size for a molecular cloud core , and coincident with the c@xmath27o - derived size @xcite .
however , as expected , the size of the molecular cloud is much larger as can be seen by the large scale map of @xcite which covers more than @xmath28 on the sky .
interestingly , due to its relatively high galactic latitude ( @xmath29 ) , this source is located about 450 pc below the mid - plane of the galactic disc .
this value is among the largest known for molecular clouds at this distance in the outer galaxy @xcite , placing this source at the edge of the galactic ( thin ) disc @xcite .
a similar large vertical distance was found for the star formation site reported by @xcite . in that case , flaring and the location of the galactic warp could account for the presence of the star formation site . however , this is not the case for iras 06345 - 3023 .
in fact , flaring of the gas component ( both hi and h@xmath30 ) in the galactic disc is known to occur but is significant only at larger galactocentric distances @xcite .
thus , the present day existence of molecular gas and dust at this relatively large vertical distance raises new questions about the evolution of the galactic disc , as well as the ability to form stars in this extreme environment . in order to determine the nature of the group of stars seen in the near - infrared images , and
specially those seen towards the nebula , photometry of the stars in the @xmath0 images was performed .
not many stars are seen , neither field stars ( background or foreground ) nor embedded young stars .
the scarcity of field stars is due to the location of this star formation site in the outer galaxy , and about 450 pc below the galactic disc . on the other hand ,
given the good resolution and depth of the images , and considering the distance , the scarcity of young stars forming in this cloud core could indicate a lack of ability to form rich clusters at this large distance below the galactic disc .
table 1 presents the results of our photometry for the sources detected in the @xmath0-bands in the region shown in fig.[rgb995 - 1 ] .
columns ( 1 ) and ( 2 ) give their coordinates .
the @xmath0 magnitudes ( in the 2mass system ) are presented in columns ( 3 ) , ( 4 ) and ( 5 ) , where errors are of the order of 0.10 mag . in column
( 6 ) we indicate the sources that are likely to be young stellar objects according to their near - infrared @xmath31 and @xmath32 colours @xcite .
the sources labeled in fig .
[ rgb999 ] are indicated by their names .
only nine stars are detected as young stellar object candidates and likely to be forming in this region .
their near - infrared colours are compatible with class i ysos , that is optically invisible objects still surrounded by cirumstellar discs and envelopes and frequently exhibiting modest outflows .
in fact , these young stars are not optically visible ( absent in the digitized sky survey ) even though they have created regions of lower extinction through the cloud that manifest as optical nebulae .
we estimate the luminosity of the brightest one ( irs 3 ) by integrating the near - infrared fluxes and the iras fluxes .
the result is 41@xmath33 and it represents an upper limit for the luminosity of the brightest object in the @xmath4-band .
this low value of the luminosity for a pre - main - sequence object implies that no high - mass star is present in this star formation site . .vlt
/ isaac photometry of sources towards iras 063453023 [ cols="^,^,^,^,^,^,^ " , ]
high - resolution , deep , near - infrared ( @xmath0 ) images of the region towards iras 06345 - 3023 reveals a small group of stars compatible with the presence of a small aggregate of young stars .
2 . the stars are located in a region exhibiting nebular emission and are embedded in a dense molecular cloud core detected through co line emission .
the nebular light is rich in morphological structures , including arcs in the vicinity of embedded stars , wisps , and bright rims of a butterfly - shaped dark cloud . in particular
, the circular arc may represent a circumbinary ring .
4 . at about 1,5 kpc ( heliocentric distance ) in the outer galaxy , and at a galactocentric distance of 10 kpc , the relatively large value of its galactic latitude places this group of young stars at 450 pc below the galactic plane , at the edge of the molecular disc . thus , active star formation is taking place at vertical distances larger than those typical of the ( thin ) disc .
this may raise questions on the origin of the gas forming this cloud and on the sharpness of the stellar and of the gaseous galactic thin disc .
the group of young stars is likely to be composed of low - mass stars , mostly class i and flat spectrum young stellar objects .
6 . inspection of wise images and catalogues confirms the yso nature of the sources and the absence of additional young stars associated to this star formation site .
this research was based on observations collected at the eso 8.2-m vlt - ut1 antu telescope ( program 68.c-0214a ) .
jy acknowledges support from fct ( sfrh / bsab/1423/2014 ) .
this research made use of the nasa/ ipac infrared science archive , which is operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration .
this research also made use of the simbad database , operated at cds , strasbourg , france , as well as saoimage ds9 , developed by the smithsonian astrophysical observatory .
this publication makes use of data products from the wide - field infrared survey explorer , which is a joint project of the university of california , los angeles , and the jet propulsion laboratory / california institute of technology , funded by the national aeronautics and space administration . | we report the discovery of a small aggregate of young stars seen in high - resolution , deep near - infrared ( @xmath0 ) images towards iras 06345 - 3023 in the outer galaxy and well below the mid - plane of the galactic disc .
the group of young stars is likely to be composed of low - mass stars , mostly class i young stellar objects .
the stars are seen towards a molecular cloud whose co map peaks at the location of the iras source .
the near - infrared images reveal , additionally , the presence of nebular emission with rich morphological features , including arcs in the vicinity of embedded stars , wisps and bright rims of a butterfly - shaped dark cloud . the location of this molecular cloud as a new star formation site well below the galactic plane in the outer galaxy indicates that active star formation is taking place at vertical distances larger than those typical of the ( thin ) disc .
[ firstpage ] stars : formation infrared : stars ism : clouds ism : individual objects : iras 06345 - 3023 dust , extinction galaxy : stellar content . |
a new branch of physics called spintronics has seen a rapid development over the last decades .
the central theme is the active manipulation of spin degrees of freedom as well as charge ones of electrons ; spintronics avoids the dissipation from joule heating by replacing charge currents with spin currents . thus establishing methods for the generation of a spin current
is significant from viewpoints of fundamental science and potential applications to green information and communication technologies.@xcite a standard way to generate a ( pure ) spin current is the spin pumping effect at the interface between a ferromagnetic material and a non - magnetic metal .
there the precession of the magnetization induces a spin current pumped into a non - magnetic metal , which is proportional to the rate of precession of the magnetic moment .
the precessing ferromagnet acts as a source of spin angular momentum ; spin battery .
this method has been developed by tserkovnyak et al.@xcite now their formula has been widely used for interpreting vast experimental results , despite the phenomenological treatment of spin - flip scattering processes .
thus we reformulate the spin pumping theory , through the schwinger - keldysh formalism , to explicitly describe the spin - flip processes .
a generation of the pumped spin current is discussed on the basis of the spin continuity equation for conduction electrons .
moreover the utilization of magnons , which are the quantized collective motions of localized spins , has recently been attracting considerable interest.@xcite hence , the focus of the present work lies also on the contribution of magnons to spin pumping .
we treat localized spins as not classical variables but magnon degrees of freedom
. therefore we can capture the ( non - equilibrium ) spin - flip dynamics , where spin angular momentum is exchanged between conduction electrons and localized spins , on the basis of the rigorous quantum mechanical theory .
consequently , we can reveal the significance of time - dependent transverse magnetic fields , which act as quantum fluctuations , for the spin pumping effect . .
conduction electrons can not enter the ferromagnet , which is an insulator .
[ fig : pumping ] , width=283 ] we consider a ferromagnetic insulator and non - magnetic metal junction@xcite shown in fig .
[ fig : pumping ] where conduction electrons couple with localized spins @xmath0 , @xmath1 , at the interface ; @xmath2 the exchange coupling constant reads @xmath3 , and the lattice constant of the ferromagnet is @xmath4 .
the magnitude of the interaction is supposed to be constant and we adopt the continuous limit in the present study .
conduction electron spin variables are represented as @xmath5/2 \\ & \equiv ( c^{\dagger } \sigma ^j c)/2 , \end{split}\ ] ] where @xmath6 are the @xmath7 pauli matrices ; @xmath8 = 2i\epsilon _ { jkl } \sigma ^l $ ] , ( @xmath9 ) .
operators @xmath10 are creation / annihilation operators for conduction electrons , which satisfy the ( fermionic ) anticommutation relation ; @xmath11 .
we focus on the dynamics at the interface where spin angular momentum is exchanged between conduction electrons and the ferromagnet .
we suppose the uniform magnetization and thus localized spin degrees of freedom can be mapped into magnon ones via the holstein - primakoff transformation ; @xmath12a({\mathbf{x}},t ) + { \cal{o}}({\tilde s}^{-3/2 } ) , \end{split}\ ] ] @xmath13 + { \cal{o}}({\tilde s}^{-3/2 } ) , \end{split}\ ] ] @xmath14 @xmath15 , where operators @xmath16 are magnon creation / annihilation operators satisfying the ( bosonic ) commutation relation ; @xmath17= \delta ( \mathbf{x}-\mathbf{x ' } ) $ ] . up to the @xmath18 terms ,
localized spins reduce to a free boson system .
consequently in the quadratic dispersion ( i.e. long wavelength ) approximation , the localized spin with the applied magnetic field along the quantization axis ( z - axis ) is described by the hamiltonian @xmath19 ; @xmath20 and the hamiltonian , @xmath21 , can be rewritten as @xmath22 c^{\dagger } ( { \mathbf{x}},t ) \sigma ^{+ } c({\mathbf{x}},t ) \nonumber \\ & + & \big[1-\frac{a^{\dagger } ( { \mathbf{x}},t ) a({\mathbf{x}},t ) } { 4\tilde s } \big ] a({\mathbf{x}},t ) c^{\dagger } ( { \mathbf{x}},t ) \sigma ^{- } c({\mathbf{x}},t ) \bigg\}. \label{eqn : e3}\end{aligned}\ ] ] the variable @xmath23 represents the effective mass of a magnon .
we have denoted a constant applied magnetic field along the quantization axis as @xmath24 , which includes @xmath25-factor and bohr magneton . in this paper , we take @xmath26 for convenience . we have adopted the continuous limit and hence , as shown in fig .
[ fig : pumping ] ( b ) , the interface is defined as an effective area where the fermi gas ( conduction electrons ) and the bose gas ( magnons ) coexist to interact .
the dynamics is described by the hamiltonian @xmath27 and the width of the interface is supposed to be of the order of the lattice constant.@xcite eq .
( [ eqn : e3 ] ) shows that localized spins at the interface lose spin angular momentum by emitting magnons and conduction electrons flip from down to up by absorbing the spin angular momentum ( see fig .
[ fig : pumping ] ) , and vice versa .
this hamiltonian @xmath27 , which describes the interchange of spin angular momentum between localized spins and conduction electrons , is essential to spin pumping mediated by magnons .
therefore we clarify the contribution of magnons accompanying this exchange interaction to spin pumping , under a time - dependent transverse magnetic field .
this is the main purpose of this paper .
we also discuss how to enhance the spin pumping effect .
this paper is structured as follows .
first , through the heisenberg equation of motion , the spin relaxation torque , which breaks the spin conservation law for conduction electrons , is defined in [ sec : source ] .
second , we evaluate it through the schwinger - keldysh formalism in [ sec : kel ] .
third , we discuss how to enhance the spin pumping effect mediated by magnons through quantum fluctuations in [ sec : pumping ] .
last , we discuss the distinction between our quantum spin pumping theory and the one proposed by tserkovnyak et al.@xcite in [ sec : tserkovnyak ] .
let us remark that in this paper , we use the term , @xmath28 , to indicate applying time - dependent transverse magnetic fields .
we apply a time - dependent transverse magnetic field , which acts as a quantum fluctuation ; @xmath29 then the total hamiltonian of the system ( interface ) , @xmath30 , is given as @xmath31 c({\mathbf{x}},t ) .
% % % % % % % % % \label{eqn : e52 } \\ % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % v_{\rm{el}}^{\gamma } & = & \frac{\gamma ( t)}{4 } \int_{{\mathbf{x}}\in \text{(interface ) } } d { \mathbf{x } } { \ } c^{\dagger } ( { \mathbf{x}},t ) ( \sigma^{+ } + \sigma^{- } ) c({\mathbf{x}},t ) .
% % % % % % % % % % % % % % \label{eqn : e53 } \\ % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % v_{\rm{mag}}^{\gamma } & = & \gamma ( t)\sqrt{\frac{\tilde s}{2 } } \int_{{\mathbf{x}}\in \text{(interface ) } } d { \mathbf{x } } { \ } \bigg\ { \big[1-\frac{a^{\dagger } ( { \mathbf{x}},t ) a({\mathbf{x}},t ) } { 4\tilde s } \big ] a({\mathbf{x}},t ) \nonumber \\ & + & a^{\dagger } ( { \mathbf{x}},t ) \big[1-\frac{a^{\dagger } ( { \mathbf{x}},t ) a({\mathbf{x}},t ) } { 4\tilde s } \big ] \bigg\}. % % % % % % % % % % % % % % % % % % % \label{eqn : e4}\end{aligned}\ ] ] the variable @xmath32 denotes the effective mass of a conduction electron . here
let us mention that in the last section , we have rewritten localized spin degrees of freedom into magnon ones via the holstein - primakoff transformation .
there , of course , off - diagonal terms ( magnon - magnon interactions ) may emerge as well as eq .
( [ eqn : e2 ] ) , but they can not satisfy the resonance condition discussed at [ subsec : pump ] in detail . consequently , their contribution to spin pumping is extremely smaller than @xmath33 ( eq . ( [ eqn : e4 ] ) ) and negligible .
therefore we omit such interactions and concentrate on the effect of @xmath33 , which represents the coupling with the time - dependent transverse magnetic field .
the spin relaxation torque ( srt),@xcite @xmath34 , is defined as the term which breaks the spin conservation law for conduction electrons;@xcite @xmath35 where the dot denotes the time derivative , @xmath36 is the spin current density , and @xmath37 represents the z - component of the spin density .
we here have defined the spin density of the system as the expectation value ( estimated for the total hamiltonian , @xmath38 ) ; @xmath39 in this paper , we focus on the z - component of the srt . through the heisenberg equation of motion , the z - component of the srt is defined as @xmath40 c^{\dagger } ( { \mathbf{x}},t ) \sigma^{+ } c({\mathbf{x}},t ) \nonumber \\ & - & \big[1-\frac { a^{\dagger } ( { \mathbf{x}},t ) a({\mathbf{x}},t)}{4 \tilde s } \big]a({\mathbf{x}},t ) c^{\dagger } ( { \mathbf{x}},t ) \sigma^{- } c ( { \mathbf{x}},t ) \big{\ } } \nonumber \\ & + & \frac{\gamma ( t)}{4i } \big [ c^{\dagger } ( { \mathbf{x}},t ) \sigma^{+ } c({\mathbf{x}},t ) - c^{\dagger } ( { \mathbf{x}},t ) \sigma^{- } c ( { \mathbf{x}},t ) \big ] \bigg{\rangle}. \label{eqn : e5 } \end{aligned}\ ] ] this term arises from @xmath41 and @xmath42 , which consist of electron spin - flip operators ; @xmath43/i .
\end{aligned}\ ] ] eq .
( [ eqn : e5 ] ) shows that the srt operates the coherent magnon state.@xcite according to ralph et al .
, @xcite the net flux of nonequilibrium spin current ( i.e. the net spin current ) pumped through the surface of the interface , @xmath44 , can be computed simply by integrating the srt over the volume of the interface ; @xmath45 in addition in our case , conduction electrons can not enter the ferromagnet , which is an insulator.@xcite therefore as shown in fig .
[ fig : pumping ] ( b ) , the z - component of the net spin current pumped into non - magnetic metal can be expressed as @xmath46 thus from now on , we focus on @xmath47 and qualitatively clarify the behavior of the spin pumping effect mediated by magnons at room temperature .
let us mention that the above relation between the srt and the pumped net spin current can be understood via the spin continuity equation , eq .
( [ eqn : continuity ] ) , and we discuss at [ subsec : conservation ] . here
let us emphasize that the spin conservation law for localized spins ( i.e. magnons ) is also broken .
the magnon continuity equation for localized spins , @xcite which corresponds to the equation of motion for localized spins@xcite and describes the dynamics , reads @xmath48 where @xmath49 is the magnon current density , and @xmath50 represents the z - component of the magnon density .
we have defined the magnon density of the system also as the expectation value ( estimated for the total hamiltonian , @xmath38 ) ; @xmath51 in addition , we call @xmath52 the magnon source term @xcite , which breaks the magnon conservation law . this term arises also from @xmath53 and @xmath33 ; @xmath54/i \\ & = & \bigg{\langle } ija_0 ^ 3 \sqrt{\frac{\tilde s}{2 } } \bigg{\ { } a^{\dagger } ( { \mathbf{x}},t ) \big[1-\frac { a^{\dagger } ( { \mathbf{x}},t ) a({\mathbf{x}},t)}{4 \tilde s } \big ] c^{\dagger } ( { \mathbf{x}},t ) \sigma^{+ } c({\mathbf{x}},t ) \nonumber \\ & - & \big[1-\frac { a^{\dagger } ( { \mathbf{x}},t ) a({\mathbf{x}},t)}{4 \tilde s } \big]a({\mathbf{x}},t ) c^{\dagger } ( { \mathbf{x}},t ) \sigma^{- } c ( { \mathbf{x}},t ) \bigg{\ } } \nonumber \\ & + & i\gamma ( t ) \sqrt{\frac{\tilde s}{2 } } \bigg\ { \big[1-\frac { a^{\dagger } ( { \mathbf{x}},t ) a({\mathbf{x}},t)}{4 \tilde s } \big]a({\mathbf{x}},t ) - a^{\dagger } ( { \mathbf{x}},t ) \big[1-\frac { a^{\dagger } ( { \mathbf{x}},t ) a({\mathbf{x}},t)}{4 \tilde s } \big ] \bigg\ } \bigg{\rangle}. \end{aligned}\ ] ] within the same approximation with the srt ( , which is discussed at the next section in detail ) , this magnon source term , in fact , satisfies the relation ; @xmath55 then , the z - component of the spin continuity equation for the total system ( i.e. conduction electrons and magnons ) becomes @xmath56 where the density of the total spin angular momentum , @xmath57 , is defined as @xmath58 and consequently the z - component of the total spin current density , @xmath59 , becomes @xmath60 ( note that , @xmath61 , via the holstein - primakoff transformation ) .
the spin continuity equation for the whole system , eq .
( [ eqn : e555 ] ) , means that though each spin conservation law for electrons and magnons is broken ( see eqs .
( [ eqn : continuity ] ) and ( [ eqn : e554 ] ) ) , the total spin angular momentum is , of course , conserved .
the interface is , in general , a weak coupling regime;@xcite the exchange interaction , @xmath62 , is supposed to be smaller than the fermi energy and the exchange interaction among ferromagnets . thus @xmath41 can be treated as a perturbative term .
in addition , we apply weak transverse magnetic fields
. then we can treat @xmath41 , @xmath63 , and @xmath64 as perturbative terms to evaluate the srt , eq .
( [ eqn : e5 ] ) . through the standard procedure of the schwinger - keldysh ( or contour - ordered ) green s function ( see also appendix [ subsec :
boson]),@xcite the langreth method ( see also appendix [ subsec : concrete]),@xcite each term of the srt can be evaluated as follows ; the first term of eq .
( [ eqn : e5 ] ) reads ( see also appendix [ subsec : ours ] ) @xmath65 c^{\dagger } ( { \mathbf{x}},t ) \sigma^{+ } c({\mathbf{x}},t ) \bigg{\rangle } { \ \ \ \ } \\
% % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % = \frac{j s}{2 } ( \frac{\gamma _ 0}{2})^2 \int \frac { d { \mathbf{k } } } { ( 2\pi)^3 } \int \frac{d\omega}{2\pi } \big[1-\frac{i}{\tilde s } \int \frac { d { { \mathbf{k}}^{\prime } } } { ( 2\pi)^3 } \int \frac{d{\omega}^{\prime } } { 2\pi } { \rm{g}}^{\rm{<}}_{{\mathbf{k^{\prime } } } , \omega^{\prime } } \big ] \\ % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % \times \bigg\ { \big [ { \rm{e}}^{2i\omega t } { \rm{g}}^{\rm{a}}_{0 , \omega } + { \rm{g}}^{\rm{a}}_{0 , -\omega } \big ] \big [ { \mathcal{g}}^{\rm{t}}_{\downarrow , { \mathbf{k } } , \omega -\omega } { \mathcal{g}}^{\rm{t}}_{\uparrow , { \mathbf{k } } , \omega } - { \mathcal{g}}^{\rm{<}}_{\downarrow , { \mathbf{k } } , \omega -\omega } { \mathcal{g}}^{\rm{>}}_{\uparrow , { \mathbf{k } } , \omega } \big ] { \ \ } \\ % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % + \big [ { \rm{e}}^{-2i\omega t } { \rm{g}}^{\rm{a}}_{0 , -\omega } + { \rm{g}}^{\rm{a}}_{0 , \omega } \big ] \big [ { \mathcal{g}}^{\rm{t}}_{\downarrow , { \mathbf{k } } , \omega + \omega } { \mathcal{g}}^{\rm{t}}_{\uparrow , { \mathbf{k } } , \omega } - { \mathcal{g}}^{\rm{<}}_{\downarrow , { \mathbf{k } } , \omega + \omega } { \mathcal{g}}^{\rm{>}}_{\uparrow , { \mathbf{k } } , \omega } \big ] \bigg\ } \\ % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % + { \cal{o}}(j^2 ) + { \cal{o}}(\gamma ^4 ) + { \cal{o}}(js^{-1 } ) . {
\ \ \ \ \ } \label{eqn : e6 } \end{split}\ ] ] the variable @xmath66 is the fermionic time - ordered ( lesser , greater ) green s function , and @xmath67 is the bosonic lesser ( advanced ) one . we here have taken the extended time ( i.e. the contour variable ) defined on the schwinger - keldysh closed time path,@xcite c , on the forward path @xmath68 ( see also fig .
a@xmath691 in our manuscript@xcite ) ; @xmath70 . even when the time is located on the backward path @xmath71 , the result of the calculation does not change because each green s function is not independent ; @xmath72 , where @xmath73 represents the retarded ( advanced ) green s function .
this relation comes into effect also for the bosonic case ( see eq .
( [ eqn : appe10 ] ) )
. here it would be useful to mention that under the thermal equilibrium condition where temperature difference does not exist between ferromagnet and non - magnetic metal , the @xmath74 term including no quantum fluctuations can not contribute to spin pumping because of the balance between thermal fluctuations in ferromagnet and those in non - magnetic metal.@xcite the second term of eq .
( [ eqn : e5 ] ) reads ( see also appendix [ subsec : ours ] ) @xmath75 a({\mathbf{x}},t ) c^{\dagger } ( { \mathbf{x}},t ) \sigma^{- } c({\mathbf{x}},t ) \bigg{\rangle } { \ \ \ } \\ % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % = - \frac{j s}{2 } ( \frac{\gamma _ 0}{2})^2 \int \frac { d { \mathbf{k } } } { ( 2\pi)^3 } \int \frac{d\omega}{2\pi } \big[1-\frac{i}{\tilde s } \int \frac { d { { \mathbf{k}}^{\prime } } } { ( 2\pi)^3 } \int \frac{d{\omega}^{\prime } } { 2\pi } { \rm{g}}^{\rm{<}}_{{\mathbf{k^{\prime } } } , \omega^{\prime } } \big ] \\
% % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % \times \bigg\ { \big [ { \rm{e}}^{2i\omega t } { \rm{g}}^{\rm{r}}_{0 , \omega } + { \rm{g}}^{\rm{r}}_{0 , -\omega } \big ] \big [ { \mathcal{g}}^{\rm{t}}_{\uparrow , { \mathbf{k } } , \omega -\omega } { \mathcal{g}}^{\rm{t}}_{\downarrow , { \mathbf{k } } , \omega } - { \mathcal{g}}^{\rm{<}}_{\uparrow , { \mathbf{k } } , \omega -\omega } { \mathcal{g}}^{\rm{>}}_{\downarrow , { \mathbf{k } } , \omega } \big ] { \ \ \ } \\ % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % + \big [ { \rm{e}}^{-2i\omega t } { \rm{g}}^{\rm{r}}_{0 , -\omega } + { \rm{g}}^{\rm{r}}_{0 , \omega } \big ] \big [ { \mathcal{g}}^{\rm{t}}_{\uparrow , { \mathbf{k } } , \omega + \omega } { \mathcal{g}}^{\rm{t}}_{\downarrow , { \mathbf{k } } , \omega } - { \mathcal{g}}^{\rm{<}}_{\uparrow , { \mathbf{k } } , \omega + \omega } { \mathcal{g}}^{\rm{>}}_{\downarrow , { \mathbf{k } } , \omega } \big ] \bigg\ } { \ } \\ % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % + { \cal{o}}(j^2 ) + { \cal{o}}(\gamma ^4)+ { \cal{o}}(js^{-1 } ) , { \ \ \ } \label{eqn : e7 } \end{split}\ ] ] where the variable @xmath76 is the bosonic retarded green s function . the third term of eq .
( [ eqn : e5 ] ) reads ( see also appendix [ subsec : ours ] ) @xmath77 \\
% % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % \times \bigg\ { ( { \rm{e}}^{2i\omega t}+1 ) { \rm{g}}^{\rm{r}}_{0 , -\omega } \big [ { \mathcal{g}}^{\rm{t}}_{\downarrow , { \mathbf{k } } , \omega -\omega } { \mathcal{g}}^{\rm{t}}_{\uparrow , { \mathbf{k } } , \omega } - { \mathcal{g}}^{\rm{<}}_{\downarrow , { \mathbf{k } } , \omega -\omega } { \mathcal{g}}^{\rm{>}}_{\uparrow , { \mathbf{k } } , \omega } \big ] { \ \ \ \ \ } \\ % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % + ( { \rm{e}}^{-2i\omega t}+1 ) { \rm{g}}^{\rm{r}}_{0 , \omega } \big [ { \mathcal{g}}^{\rm{t}}_{\downarrow , { \mathbf{k } } , \omega + \omega } { \mathcal{g}}^{\rm{t}}_{\uparrow , { \mathbf{k } } , \omega } - { \mathcal{g}}^{\rm{<}}_{\downarrow , { \mathbf{k } } , \omega + \omega } { \mathcal{g}}^{\rm{>}}_{\uparrow , { \mathbf{k } } , \omega } \big ] \bigg\ } { \ \ \ \ \ } \\ % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % + { \cal{o}}(j^0 ) + { \cal{o}}(j^2 ) + { \cal{o}}(\gamma ^4)+ { \cal{o}}(js^{-1 } ) . {
\ \ \ \ \ } \label{eqn : e8 } \end{split}\ ] ] we omit the @xmath78 terms because they contain no contributions of magnons via the exchange interaction and are not relevant to the spin pumping effect ; the terms are out of the purpose of the present study . the last term of eq .
( [ eqn : e5 ] ) reads ( see also appendix [ subsec : ours ] ) @xmath79 \\
% % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % \times \bigg\ { ( { \rm{e}}^{2i\omega t}+1 ) { \rm{g}}^{\rm{a}}_{0 , -\omega } \big [ { \mathcal{g}}^{\rm{t}}_{\uparrow , { \mathbf{k } } , \omega -\omega } { \mathcal{g}}^{\rm{t}}_{\downarrow , { \mathbf{k } } , \omega } - { \mathcal{g}}^{\rm{<}}_{\uparrow , { \mathbf{k } } , \omega -\omega } { \mathcal{g}}^{\rm{>}}_{\downarrow , { \mathbf{k } } , \omega } \big ] { \ \ \ \ \ } \\ % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % + ( { \rm{e}}^{-2i\omega t}+1 ) { \rm{g}}^{\rm{a}}_{0 , \omega } \big [ { \mathcal{g}}^{\rm{t}}_{\uparrow , { \mathbf{k } } , \omega + \omega } { \mathcal{g}}^{\rm{t}}_{\downarrow , { \mathbf{k } } , \omega } - { \mathcal{g}}^{\rm{<}}_{\uparrow , { \mathbf{k } } , \omega + \omega } { \mathcal{g}}^{\rm{>}}_{\downarrow , { \mathbf{k } } , \omega } \big ] \bigg\ } { \ \ \ \ \ } \\ % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % + { \cal{o}}(j^0 ) + { \cal{o}}(j^2 ) + { \cal{o}}(\gamma ^4)+ { \cal{o}}(js^{-1 } ) . {
\ \ \ \ \ } \label{eqn : e9 } \end{split}\ ] ] finally , the srt can be rearranged as @xmath80 where @xmath81 \nonumber \\ & \times & \big[({\rm{g}}^{\rm{a}}_{0 , \omega } + { \rm{g}}^{\rm{r}}_{0 , -\omega } ) ( { \mathcal{g}}^{\rm{t}}_{\downarrow , { \mathbf{k } } , \omega -\omega } { \mathcal{g}}^{\rm{t}}_{\uparrow , { \mathbf{k } } , \omega } - { \mathcal{g}}^{\rm{<}}_{\downarrow , { \mathbf{k } } , \omega -\omega } { \mathcal{g}}^{\rm{>}}_{\uparrow , { \mathbf{k } } , \omega } ) \nonumber \\ & - & ( { \rm{g}}^{\rm{r}}_{0 , \omega } + { \rm{g}}^{\rm{a}}_{0 , -\omega } ) ( { \mathcal{g}}^{\rm{t}}_{\uparrow , { \mathbf{k } } , \omega -\omega } { \mathcal{g}}^{\rm{t}}_{\downarrow , { \mathbf{k } } , \omega } - { \mathcal{g}}^{\rm{<}}_{\uparrow , { \mathbf{k } } , \omega -\omega } { \mathcal{g}}^{\rm{>}}_{\downarrow , { \mathbf{k } } , \omega } ) \big ] , \label{eqn : e71 } \end{aligned}\ ] ] @xmath82 \nonumber \\ & \times & \big [ ( { \rm{g}}^{\rm{a}}_{0 , -\omega } + { \rm{g}}^{\rm{r}}_{0 , \omega } ) ( { \mathcal{g}}^{\rm{t}}_{\downarrow , { \mathbf{k } } , \omega + \omega } { \mathcal{g}}^{\rm{t}}_{\uparrow , { \mathbf{k } } , \omega } - { \mathcal{g}}^{\rm{<}}_{\downarrow , { \mathbf{k } } , \omega + \omega } { \mathcal{g}}^{\rm{>}}_{\uparrow , { \mathbf{k } } , \omega } ) \nonumber \\ & - & ( { \rm{g}}^{\rm{r}}_{0 , -\omega } + { \rm{g}}^{\rm{a}}_{0 , \omega } ) ( { \mathcal{g}}^{\rm{t}}_{\uparrow , { \mathbf{k } } , \omega + \omega } { \mathcal{g}}^{\rm{t}}_{\downarrow , { \mathbf{k } } , \omega } - { \mathcal{g}}^{\rm{<}}_{\uparrow , { \mathbf{k } } , \omega + \omega } { \mathcal{g}}^{\rm{>}}_{\downarrow , { \mathbf{k } } , \omega } ) \big ] , \label{eqn : e72 } \end{aligned}\ ] ] @xmath83 \nonumber \\ & \times & \big [ ( { \rm{g}}^{\rm{a}}_{0 , -\omega } + { \rm{g}}^{\rm{r}}_{0 , -\omega } ) ( { \mathcal{g}}^{\rm{t}}_{\downarrow , { \mathbf{k } } , \omega -\omega } { \mathcal{g}}^{\rm{t}}_{\uparrow , { \mathbf{k } } , \omega } - { \mathcal{g}}^{\rm{<}}_{\downarrow , { \mathbf{k } } , \omega -\omega } { \mathcal{g}}^{\rm{>}}_{\uparrow , { \mathbf{k } } , \omega } ) \nonumber \\ & + & ( { \rm{g}}^{\rm{a}}_{0 , \omega } + { \rm{g}}^{\rm{r}}_{0 , \omega } ) ( { \mathcal{g}}^{\rm{t}}_{\downarrow , { \mathbf{k } } , \omega + \omega } { \mathcal{g}}^{\rm{t}}_{\uparrow , { \mathbf{k } } , \omega } - { \mathcal{g}}^{\rm{<}}_{\downarrow , { \mathbf{k } } , \omega + \omega } { \mathcal{g}}^{\rm{>}}_{\uparrow , { \mathbf{k } } , \omega } ) \nonumber \\ & - & ( { \rm{g}}^{\rm{r}}_{0 , -\omega } + { \rm{g}}^{\rm{a}}_{0 , -\omega } ) ( { \mathcal{g}}^{\rm{t}}_{\uparrow , { \mathbf{k } } , \omega -\omega } { \mathcal{g}}^{\rm{t}}_{\downarrow , { \mathbf{k } } , \omega } - { \mathcal{g}}^{\rm{<}}_{\uparrow , { \mathbf{k } } , \omega -\omega } { \mathcal{g}}^{\rm{>}}_{\downarrow , { \mathbf{k } } , \omega } ) \nonumber \\ & - & ( { \rm{g}}^{\rm{r}}_{0 , \omega } + { \rm{g}}^{\rm{a}}_{0 , \omega } ) ( { \mathcal{g}}^{\rm{t}}_{\uparrow , { \mathbf{k } } , \omega + \omega } { \mathcal{g}}^{\rm{t}}_{\downarrow , { \mathbf{k } } , \omega } - { \mathcal{g}}^{\rm{<}}_{\uparrow , { \mathbf{k } } , \omega + \omega } { \mathcal{g}}^{\rm{>}}_{\downarrow , { \mathbf{k } } , \omega } ) \big ] .
\label{eqn : e21 } \end{aligned}\ ] ] here let us mention that , in real materials , there does exist impurity scattering .
we assume that this is the main cause for the finite lifetime of magnons and conduction electrons .
moreover the rate of impurities such as lattice defects and nonmagnetic impurities is , in general , far larger than that of magnetic impurities .
therefore we phenomenologically introduce the lifetime and regard it as a constant parameter .
then we adopt green s functions including the the effects of impurities as the lifetime , and calculate the srt by using them ; eqs .
( [ eqn : g1])-([eqn : g8 ] ) .
though , as the result , it might be better to execute the accompanying vertex corrections from viewpoints of theoretical aspects , we have not done for the aim now explained ; to put it briefly , in order to clarify that pumped spin currents are generated purely by quantum fluctuations , we have not executed vertex corrections .
it should be noted that , before our present study , takeuchi et al.@xcite have already studied spin pumping , on the basis of schwinger - keldysh formalism , under the same condition with ours except two points ; ( i ) they have treated localized spins as not magnons but classical variables , and ( ii ) they have not applied any transverse magnetic fields . on their condition , they have clarified that , under the uniform magnetization , spin currents can not be generated without vertex corrections ( i.e. multiple scatterings of impurities ) . in other words , they have already revealed that spin currents can be generated by the effects of multiple scatterings of impurities , i.e. vertex corrections .
thus , the main purpose of the present study is to propose an alternative way for the generation of spin currents without using vertex corrections , i.e. multiple scatterings of impurities ; we propose a method for the generation of spin currents by using time - dependent transverse magnetic fields , which are under our control and act as quantum fluctuations .
therefore we call this method quantum spin pumping . in order to clarify
that pumped spin currents are induced purely by quantum fluctuations , we have not included the effects of multiple scatterings of impurities ( i.e. we have not executed vertex corrections ) .
each green s function including the the effects of impurities as the lifetime reads as follows ( see also appendix [ subsec : boson]);@xcite @xmath84^{-1 } \label{eqn : g1 } \\ & = & ( { \rm{g}}^{\rm{r } } _ { { \mathbf{k^{\prime } } } , \omega^{\prime } } ) ^{\ast } , \label{eqn : g2 } \\ % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % { \mathcal{g}}_{\sigma , { \mathbf{k } } , \omega}^{\rm{a } } & = & [ \omega - \omega_{\sigma , \mathbf{k } } - i/(2\tau ) ] ^{-1 } \label{eqn : g3 } \\ & = & ( { \mathcal{g}}_{\sigma , { \mathbf{k } } , \omega}^{\rm{r}})^{\ast } , \label{eqn : g4 } \\ % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % { \rm{g}}^{\rm { < } } _ { { \mathbf{k^{\prime } } } , \omega^{\prime } } & = & -f_{\rm{b}}(\omega^{\prime } ) ( { \rm{g}}^{\rm{a } } _ { { \mathbf{k^{\prime } } } , \omega^{\prime } } - { \rm{g}}^{\rm{r } } _ { { \mathbf{k^{\prime } } } , \omega^{\prime } } ) , \label{eqn : g5 } \\ % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % { \mathcal{g}}_{\sigma , { \mathbf{k } } , \omega}^{\rm { < } } & = & f_{\rm{f } } ( \omega ) ( { \mathcal{g}}_{\sigma , { \mathbf{k } } , \omega}^{\rm{a } } - { \mathcal{g}}_{\sigma , { \mathbf{k } } , \omega}^{\rm{r } } ) , \label{eqn : g6 } \\ % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % { \mathcal{g}}_{\sigma , { \mathbf{k } } , \omega}^{\rm{t } } & = & { \mathcal{g}}_{\sigma , { \mathbf{k } } , \omega}^{\rm{r } } + { \mathcal{g}}_{\sigma , { \mathbf{k } } , \omega}^{\rm { < } } \label{eqn : g7 } \\ & = & { \mathcal{g}}_{\sigma , { \mathbf{k } } , \omega}^{\rm { > } } + { \mathcal{g}}_{\sigma , { \mathbf{k } } , \omega}^{\rm{a } } , \label{eqn : g8 } \end{aligned}\ ] ] where the variables @xmath85 , @xmath86 , @xmath87 , and @xmath88 are the lifetime of magnons , that of conduction electrons , the bose distribution function , and the fermi one , respectively .
the energy dispersion relation reads @xmath89 and @xmath90 , where @xmath91 , @xmath92 , @xmath93 , and @xmath94 denotes the chemical potential .
we consider a weak magnetic field regime and omit the @xmath95 ^ 2 ) $ ] terms , where @xmath96 represents the fermi energy . through the sommerfeld expansion ,
the chemical potential is determined as , @xmath97 .
the relation between the srt and the pumped net spin current in [ subsec : def ] , @xmath98 , can be understood via the spin continuity equation ; @xmath99 . through the same procedure and the same approximation with the last section ,
the time derivative of the spin density is estimated ( on the condition mentioned at [ subsec : pump ] ) as , @xmath100 .
thus @xmath101 in the spin continuity equation is negligible@xcite also in our case .
consequently , the spin continuity equation becomes @xmath102 .
therefore by integrating over the volume of the interface , the relation between the srt and the pumped net spin current mediated by magnons is clarified ; @xmath103 in addition in our case , conduction electrons can not enter the ferromagnet , which is an insulator .
then the net spin current pumped into the non - magnetic metal can be calculated by integrating the srt ( eq .
( [ eqn : e30 ] ) ) over the interface ( see fig . [
fig : pumping ] ) .
this is our spin pumping theory mediated by magnons via the exchange interaction at the interface between a non - magnetic metal and a ferromagnetic insulator .
the breaking of the spin conservation law for conduction electrons , @xmath47 , is essential to our spin pumping theory .
our calculation ( eqs . ( [ eqn : e20])-([eqn : e21 ] ) ) gives , @xmath104 thus our formalism ( eq . ( [ eqn : e30 ] ) ) shows that spin currents mediated by magnons can not be pumped without quantum fluctuations ; this is the significant difference from the theory proposed by tserkovnyak et al.@xcite , which is discussed at [ sec : tserkovnyak ] in detail .
that is , quantum fluctuations are essential to spin pumping mediated by magnons as well as the exchange interaction between conduction electrons and ferromagnets ( see eqs .
( [ eqn : e20])-([eqn : e21 ] ) ) ; @xmath105 this is the main feature of our quantum spin pumping theory mediated by magnons . in other words ,
we have shown that magnons accompanying the exchange interaction can not contribute to spin pumping without quantum fluctuations . .
a sharp peak exists on the point where the resonance condition , @xmath106 , is satisfied .
each quantity , @xmath107 , has the same structure with @xmath108 .
( b ) the time evolution of the srt , @xmath109 , at the resonance point ( @xmath110 ) ; ( i ) @xmath111 , ( ii ) @xmath112 , ( iii ) @xmath113 , where @xmath114 @xmath115 .
[ fig : fourier ] , width=453 ] the srt ( eq .
( [ eqn : e20 ] ) ) , @xmath116 , is rewritten as ( see also appendix [ sec : srt ] ) @xmath117 the variable , @xmath118 , is the dimensionless srt as a function of the wavenumber for conduction electrons @xmath119 , the angular frequency of the applied transverse magnetic field @xmath120 , the magnitude of the exchange interaction @xmath121 , and so on .
[ fig : fourier ] ( a ) represents the @xmath122 which is the time average of @xmath123 .
[ fig : fourier ] ( b ) shows @xmath124 ( see eq .
( [ eqn : e01 ] ) ) , which describes the time evolution of the srt .
we have set each parameter , as a typical case,@xcite as follows ; @xmath125 ev , @xmath126 , @xmath127 k , @xmath128 ev @xmath129 , @xmath130 ev @xmath129 , @xmath131 ps @xmath132 , @xmath133 , @xmath134 . fig .
[ fig : fourier ] ( a ) shows that the srt has a sharp peak as a result of the resonance with the angular frequency , @xmath135 , of the applied transverse magnetic field .
the sharp peak exists on the point where the condition , @xmath136 is satisfied .
this is because according to eq .
( [ eqn : e51 ] ) , the localized spin acts as an effective magnetic field along the quantization axis @xmath62 , which is far larger than the applied magnetic field @xmath24 ; @xmath137 this fact ( i.e. resonance condition ) is useful to enhance the spin pumping effect because the angular frequency of a transverse magnetic field is under our control .
in addition fig .
[ fig : fourier ] ( a ) also shows that the stronger the exchange interaction becomes , the larger the srt does .
[ fig : fourier ] ( b ) shows @xmath109 at the resonance point ( @xmath110 ) .
each period reads @xmath138
last , let us discuss the distinction between our quantum spin pumping theory@xcite and the one proposed by tserkovnyak et al.@xcite it should be noted that , as has been mentioned in their article@xcite ( see viii .
summary and outlook in their article@xcite ) , they have phenomenologically treated the spin - flip scattering processes , which we have regarded as the most important processes for spin pumping .
nevertheless , now their spin pumping theory has been widely used for interpreting vast experimental results,@xcite in particular by experimentalists .
thus , it would be significant to clarify the difference between our quantum spin pumping theory and the one proposed by tserkovnyak et al . according to the phenomenological@xcite spin pumping theory by tserkovnyak et al .
and their notation,@xcite the pumped spin current @xmath139 reads @xmath140 where the dot denotes the time derivative .
we have taken @xmath141 , and @xmath142 denotes a unit vector along the magnetization direction ; they have treated @xmath142 as classical variables .
the variable @xmath143 is the complex - valued mixing conductance that depends on the material;@xcite @xmath144 .
the magnetization dynamics of ferromagnets can be described by the landau - lifshitz - gilbert ( llg ) eq . ; @xmath145 where @xmath146 is the gyro - magnetic ratio and @xmath147 is the gilbert damping constant that determines the magnetization dissipation rate .
here it should be emphasized that though this gilbert damping constant , @xmath147 , was phenomenologically introduced,@xcite it can be derived microscopically by considering a whole system including spin relaxation;@xcite thus the effect of the exchange coupling to conduction electrons should be considered to have already been included into this gilbert damping term .
the effective magnetic field is set as @xmath148 where @xmath149 represents a time - dependent transverse magnetic field .
the llg eq . , eq . ( [ eqn:2 ] ) , becomes @xmath150 eq . ( [ eqn:2 - 2 ] ) is substituted into @xmath151 , eq .
( [ eqn:1 ] ) ; we include the contribution of the gilbert damping term , which depends on the materials , up to @xmath152 ; @xmath153 for @xmath154 ( metal),@xcite and @xmath155 for @xmath156 ( insulator),@xcite as examples .
their theory is applicable to both ferromagnetic metals and insulators.@xcite consequently , the z - component of the pumped spin current reads @xmath157 - \gamma \gamma { \rm{m}}^x { \rm{m}}^z - \alpha \gamma \gamma [ { \rm{m}}^x { \rm{m}}^y { \rm{m}}^x + ( { \rm{m}}^y)^3 + ( { \rm{m}}^z)^2 { \rm{m}}^y ] \big\ } \nonumber \\ & + & { \rm{g}}_{\perp } ^{(\rm{i } ) } \big\ { \alpha \gamma \ { b [ ( { \rm{m}}^x)^2 + ( { \rm{m}}^y)^2 ] - \gamma { \rm{m}}^x { \rm{m}}^z \ } + \gamma \gamma { \rm{m}}^y \big\ } + { \cal{o}}(\alpha ^2 ) .
\\ % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % & \stackrel{\gamma \rightarrow 0}{\longrightarrow } & [ { \rm{g}}_{\perp } ^{(\rm{r } ) } + \alpha { \rm{g}}_{\perp } ^{(\rm{i } ) } ] \gamma b [ ( { \rm{m}}^x)^2 + ( { \rm{m}}^y)^2 ] .
\label{eqn:3 } \\ % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % & \stackrel{{\rm{g}}_{\perp } ^{(\rm{i } ) } \rightarrow 0}{\longrightarrow } & { \rm{g}}_{\perp } ^{(\rm{r } ) } \gamma b [ ( { \rm{m}}^x)^2 + ( { \rm{m}}^y)^2 ] . \label{eqn:3 - 2 } \end{aligned}\ ] ] at finite temperature , the magnetization is thermally activated ; @xmath158.@xcite then the time derivative of the z - component means @xmath159 -\gamma { \rm{m}}^x { \rm{m}}^z\ } + { \cal{o}}(\alpha ^2 ) .
\label{eqn : e4 - 4 } \\ & \stackrel{\gamma \rightarrow 0}{\longrightarrow } & \alpha \gamma b [ ( { \rm{m}}^x)^2 + ( { \rm{m}}^y)^2 ] .
\label{eqn : e4 } \\ & \not= & 0 .
\label{eqn:4 - 3 } \end{aligned}\ ] ] eqs .
( [ eqn:3 ] ) , ( [ eqn:3 - 2 ] ) and ( [ eqn:4 - 3 ] ) mean that , within the framework by tserkovnyak et al . with the llg eq .
, they may gain spin currents at finite temperature if only the magnetic field along the z - axis , @xmath24 , is applied ; @xmath160 that is , the spin pumping theory by tserkovnyak et al.@xcite with the llg eq .
concludes that spin currents may be pumped at finite temperature without time - dependent transverse magnetic fields . on the other hand , as discussed in the last section ( see eqs .
( [ eqn : add1 ] ) and ( [ eqn : add2 ] ) ) , our approach based on the schwinger - keldysh formalism gives different result ; @xmath161 that is , our quantum spin pumping theory means that spin currents mediated by magnons can not be pumped without quantum fluctuations ( i.e. time - dependent transverse magnetic fields , @xmath149 ) ; quantum fluctuations are essential to spin pumping mediated by magnons as well as the exchange interaction between conduction electrons and ferromagnets .
this ( eqs .
( [ eqn : add3 ] ) and ( [ eqn : add4 ] ) ) is the significant distinction between our quantum spin pumping theory and the one proposed by tserkovnyak et al .
we have microscopically studied quantum spin pumping mediated by magnons by evaluating the srt at room temperature .
localized spins of the ferromagnetic insulator lose spin angular momentum by emitting a magnon and conduction electrons flip from down to up by absorbing the momentum .
thus our formalism contains no phenomenological treatments of spin - flip scattering processes .
the srt breaking the spin conservation law represents the ( net ) spin current mediated by magnons , which is pumped into the adjacent non - magnetic metal . through the schwinger - keldysh formalism
, we have clarified that quantum fluctuations ( i.e. time - dependent transverse magnetic fields ) induce a net spin current , which can be enhanced through a resonance structure as a function of the angular frequency of the applied transverse field .
we conclude that the breaking of the spin conservation law and quantum fluctuations are essential to quantum spin pumping mediated by magnons accompanying the exchange interaction . in this paper ,
we have theoretically introduced and defined the interface as an effective area where the fermi gas ( conduction electrons ) and the bose gas ( magnons ) coexist to interact . though the behavior of the quantum spin pumping effect mediated by magnons can be qualitatively captured by calculating the srt , the theoretical estimation for the volume , in particular the width , of the interface is essential for the quantitative understanding .
of course the width of the interface may be roughly supposed to be of the order of the lattice constant , but we consider microscopic derivation ( so called proximity effects ) is an urgent theoretical issue .
in addition , though we have not executed vertex corrections at the present study in order to clarify that pumped spin currents are generated purely by quantum fluctuations , calculating the srt under the effects of vertex corrections ( i.e. multiple scatterings of impurities ) as well as quantum fluctuations is a significant theoretical issue . also from viewpoints of theoretical aspects ( i.e. quantum field theory )
, it might be better to execute vertex corrections .
thus , we would like to tackle this issue in the near future .
in the present study , we have focused exclusively on the dynamics at the interface ( @xmath162 ) , where pumped spin currents are generated with interchanging spin angular momentum between conduction electrons and magnons .
then , our quantum spin pumping theory microscopically well describes the dynamics of generating pumped spin currents , in particular , the spin - flip scattering processes , which we have regarded as the most important processes for spin pumping .
this is the strong point of our theory ; note that the theory proposed by tserkovnyak et al . has phenomenologically treated spin - flip processes.@xcite on the other hand , the dynamics of pumped spin currents in the non - magnetic metal ( @xmath163 ) , i.e. how the pumped spin currents flow in the non - magnetic metal , is out of the application ( purpose ) of the present our theory .
that is , though our formalism has microscopically captured the spin - flip scattering processes at the interface beyond phenomenology , it does not cover the dynamics of pumped spin currents ; the dynamics under the existence of pumped spin currents such as the effect of localized spins and conduction electrons on pumped spin currents , which is often indicated as the @xmath164 term,@xcite is out of the application ( purpose ) of the present study .
therefore , we would like to brush up our quantum spin pumping theory to cover the dynamics of pumped spin currents in the non - magnetic metal ; how the pumped spin currents flow in the non - magnetic metal .
we recognize that clarifying these issues is significant from viewpoints of applications as well as fundamental science .
last , we have revealed that magnons accompanying the exchange interaction can not contribute to quantum spin pumping without quantum fluctuations . here
it should be stressed that in our formalism , the meditation of spin angular momentum is restricted to only magnons .
we will take the effect of phonons into account and develop our theory as a more rigorous formalism to reveal the microscopic ( quantum ) dynamics of the magnon splitting.@xcite moreover , we are also interested in the contribution of magnons to quantum spin pumping under a spatially nonuniform magnetization .
we would like to thank k. totsuka for stimulating the study , and g. tatara for reading the manuscript and useful comments .
we are also grateful to t. takahashi and y. korai for fruitful discussion , n. sago for helpful correspondence on numerical calculations , and k. ando for sending the invaluable presentation file , prior to the publication , on the 6th international school and conference on spintronics and quantum information technology ( _ spintech6 _ ) .
we also would like to thank s. onoda and t. oka for crucial comments at the 26th nishinomiya - yukawa memorial international workshop _ novel quantum states in condensed matter _ ( _ nqs2011 _ ) .
we are supported by the grant - in - aid for the global coe program `` the next generation of physics , spun from universality and emergence '' from the ministry of education , culture , sports , science and technology ( mext ) of japan .
in this section , we show how to calculate eqs .
( [ eqn : e6])-([eqn : e9 ] ) in detail .
first , we briefly show the langreth method,@xcite which is useful to evaluate the perturbation expansion of the keldysh ( or contour - ordered ) green s function@xcite in subsection [ subsec : concrete ] .
second , we introduce the point of the bosonic keldysh green s functions@xcite in subsection [ subsec : boson ] .
last , the detailed calculations of eqs .
( [ eqn : e6])-([eqn : e9 ] ) are represented in subsection [ subsec : ours ] .
we omit the label , @xmath165 , when it is not relevant . for simplicity here , we consider the perturbative term , @xmath166 , @xmath167 , \label{eqn : appe2}\end{aligned}\ ] ] and evaluate the expectation value of the bosonic annihilate operator , @xmath168 , as an example ; @xmath169 \big{\rangle } \\ & = & -i\int d{\mathbf{x}}^{\prime } \int_{\text{c } } d\tau ' \gamma ( \tau ' ) \langle \text{t}_{\text{c } } \ a(\mathbf{x } , \tau)a^{\dagger } ( \mathbf{x ' } , \tau ' ) \rangle + { \cal{o}}(\gamma ^2)\\ & \equiv & -i \int d{\mathbf{x}}^{\prime } { \cal{i}}.\ ; \label{eqn : appe3}\end{aligned}\ ] ] here @xmath170 is the path - ordering operator defined on the schwinger - keldysh closed time path,@xcite @xmath171 ( see fig .
a@xmath691 in our manuscript@xcite ) we express the schwinger - keldysh closed time path as a sum of the forward path , @xmath172 , and the backward path , @xmath173 ; @xmath174.@xcite we take @xmath175 which denotes the contour variable defined on the schwinger - keldysh closed time path on forward path , @xmath176 . even when @xmath175 is located on backward path , @xmath177 , the result of this calculation is invariant because each green s function , @xmath178 , @xmath179 , @xmath180 , @xmath181 , is not independent;@xcite they obey , @xmath182 note that this relation comes into effect also for the fermionic case;@xcite @xmath183 the integral on the schwinger - keldysh closed time path of eq .
( [ eqn : appe3 ] ) , @xmath184 , is executed by taking an identity@xcite into account @xmath185 as ( see also eq .
( [ eqn : kel ] ) in appendix [ subsec : boson ] ) @xmath186 . \ ; \end{split } \label{eqn : appe5}\ ] ] by using the relation , @xmath187 , we obtain @xmath188 in this subsection , we show the point of the bosonic keldysh green s functions . the bosonic keldysh green s function , @xmath189 , is defined as@xcite @xmath190 depending on the points where @xmath86 and @xmath191 are located on the schwinger - keldysh closed time path ( i.e. @xmath192 ) , the bosonic keldysh green s function is expressed as@xcite @xmath193 it should be noted that each green s function is not independent;@xcite @xmath194 in addition , these relations@xcite would be useful on calculation ; @xmath195 by executing the fourier transformation , the lesser and greater green s functions for free bosons become@xcite @xmath196 \delta ( \omega^{\prime}- \omega _ { { \mathbf{k^{\prime } } } } ) , \\ % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % { \rm{g}}^{\rm{k } } _ { { \mathbf{k^{\prime } } } , \omega^{\prime } } & : = & { \rm{g}}^{\rm { < } } _ { { \mathbf{k^{\prime } } } , \omega^{\prime } } + { \rm{g}}^{\rm { > } } _ { { \mathbf{k^{\prime } } } , \omega^{\prime } } \\ & = & -2 \pi i [ 1 + 2f_{\rm{b}}(\omega^{\prime } ) ] \delta ( \omega^{\prime}- \omega _ { { \mathbf{k^{\prime } } } } ) \\ & = & 2 i { \rm{im } } { \rm{g}}^{\rm{r } } _ { { \mathbf{k^{\prime } } } , \omega^{\prime } } { \rm{coth}}(\beta \omega^{\prime } /2).\end{aligned}\ ] ] the last one , @xmath197 , represents the keldysh green s function @xcite and the relation is called the bosonic fluctuation - dissipation theorem.@xcite _ fermionic keldysh green s function _
it would be useful to compare with the ( spinless ) fermionic keldysh green s function , @xmath198 , which is defined as@xcite @xmath199 depending on the points where @xmath86 and @xmath191 are located on the schwinger - keldysh closed time path ( i.e. @xmath192 ) , the fermionic keldysh green s function is expressed as@xcite @xmath200 note that with reflecting the statistical properties , the sign of the lesser green s function is opposite from the bosonic case . in addition , they satisfy the relation;@xcite @xmath201 @xmath202 by executing the fourier transformation , the lesser and greater green s functions for free fermions become@xcite @xmath203 \delta ( \omega- \omega _ { { \mathbf{k } } } ) , \\ % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % { \mathcal{g}}^{\rm{k } } _ { { \mathbf{k } } , \omega } & : = & { \mathcal{g}}^{\rm { < } } _ { { \mathbf{k } } , \omega } + { \mathcal{g}}^{\rm { > } } _ { { \mathbf{k } } , \omega } \\ & = & -2 \pi i [ 1 - 2f_{\rm{f}}(\omega ) ] \delta ( \omega - \omega _ { { \mathbf{k } } } ) \\ & = & 2 i { \rm{im } } { \mathcal{g}}^{\rm{r } } _ { { \mathbf{k } } , \omega } { \rm{tanh}}(\beta \omega /2).\end{aligned}\ ] ] the last one , @xmath204 , represents the keldysh green s function @xcite and the relation is called the fermionic fluctuation - dissipation theorem.@xcite in this subsection , we will briefly show how to calculate eqs .
( [ eqn : e6])-([eqn : e9 ] ) . by adopting the wick s theorem,@xcite the left - hand side ( lhs ) of eq .
( [ eqn : e6 ] ) reads @xmath205 by employing the relation ; @xmath206 and the langreth method@xcite with eqs . (
[ eqn : kel ] ) and ( [ eqn : kelfermion ] ) , the right - hand side ( rhs ) of eq .
( [ eqn : a10 ] ) can be expressed as @xmath207 [ { \cal{g}}_{\downarrow } ^{\rm{t}}(t , t^{\prime } ) { \cal{g}}_{\uparrow } ^{\rm{t}}(t^{\prime } , t ) - { \cal{g}}_{\downarrow } ^{\rm{<}}(t , t^{\prime } ) { \cal{g}}_{\uparrow } ^{\rm{>}}(t^{\prime } , t ) ] .
\label{eqn : a11 } \end{split}\ ] ] we also here have taken @xmath175 on forward path , @xmath176 . as discussed in the last subsection , even when @xmath175 is located on backward path ,
@xmath177 , the result of this calculation is invariant . here
it should be noted that @xmath208 then , eq . ( [ eqn : a11 ] ) can be rewritten as @xmath209 .
\label{eqn : a12 } \end{split}\ ] ] by executing fourier transformation , we obtain the rhs of eq .
( [ eqn : e6 ] ) . through the same procedure , remained terms are evaluated as follows ; @xmath210 .
\label{eqn : aaa } \end{split}\ ] ] @xmath211 .
\label{eqn : aaa2 } \end{split}\ ] ] note that we have adopted the relation ; @xmath212 . @xmath213 .
\label{eqn : aaa3 } \end{split}\ ] ] note that we have employed the relation ; @xmath214 . by using fourier transformation
, we obtain the rhs of eqs .
( [ eqn : e7])-([eqn : e9 ] ) .
finally , let us write down the explicit form of the srt . we have taken @xmath26 and have introduced the dimensionless variables as follows ; @xmath215 , except @xmath216 and @xmath217 .
the dimensionless wavenumber for conduction electrons has been rewritten ; @xmath218 . the srt ( i.e. eq . ( [ eqn : e01 ] ) ) , @xmath219 , reads @xmath220 where @xmath221 c^{\dagger } ( { \mathbf{x}},t ) \sigma^{+ } c({\mathbf{x}},t ) \bigg{\rangle } + \bigg{\langle } \frac{\gamma ( t ) c^{\dagger } ( { \mathbf{x}},t ) \sigma^{+ } c({\mathbf{x}},t ) } { 4i } \bigg{\rangle } \\ & = & \frac { js } { 4 \pi } ( \frac{\gamma _ 0}{2})^2 \int \frac{d { { \mathbf{k}}}}{(2\pi ) ^3 } \int d\omega \big[1-\frac{i}{\tilde s } \int \frac{d { { \mathbf{k}}}^{\prime}}{(2\pi ) ^3 } \int \frac{d{\omega}^{\prime } } { 2\pi } { \rm{g}}^{\rm { < } } _ { { { \mathbf{k}}}^{\prime } , \omega ^{\prime } } \big ] \nonumber \\ & \times & \bigg{\ { } % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % \big [ { \rm{e}}^{2i\omega t } { \rm{g}}^{\rm{a}}_{0 , \omega } + { \rm{g}}^{\rm{a}}_{0 , -\omega } + ( { \rm{e}}^{2i\omega t}+1 ) { \rm{g}}^{\rm{r}}_{0 , -\omega } \big ] \cdot \big[{\mathcal{g}}^{\rm{t}}_{\downarrow , { \mathbf{k } } , \omega -\omega } { \mathcal{g}}^{\rm{t}}_{\uparrow , { \mathbf{k } } , \omega } - { \mathcal{g}}^{\rm{<}}_{\downarrow , { \mathbf{k } } , \omega -\omega } { \mathcal{g}}^{\rm{>}}_{\uparrow , { \mathbf{k } } , \omega } \big ] \nonumber \\ % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % & + & \big [ { \rm{e}}^{-2i\omega t } { \rm{g}}^{\rm{a}}_{0 , -\omega } + { \rm{g}}^{\rm{a}}_{0 , \omega } + ( { \rm{e}}^{-2i\omega t}+1 ) { \rm{g}}^{\rm{r}}_{0 , \omega } \big ] \cdot \big[{\mathcal{g}}^{\rm{t}}_{\downarrow , { \mathbf{k } } , \omega + \omega } { \mathcal{g}}^{\rm{t}}_{\uparrow , { \mathbf{k } } , \omega } - { \mathcal{g}}^{\rm{<}}_{\downarrow , { \mathbf{k } } , \omega + \omega } { \mathcal{g}}^{\rm{>}}_{\uparrow , { \mathbf{k } } , \omega } \big ] % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % \bigg{\ } } \nonumber \\ & + & { \cal{o}}(j^2 ) + { \cal{o}}({(\gamma_0)}^0 ) \\ % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % & = & { \bar{j } } \int d\bar { \omega } \int_0^{\infty } dy y^2 \bigg [ \frac { { \rm{e}}^{2i\omega t } ( \bar { \omega } -\bar{b}+\frac{i}{2\bar{\tau}_{\rm{m } } } ) } { ( \bar { \omega } -\bar{b})^2 + ( \frac{1}{2\bar{\tau}_{\rm{m}}})^2 } - \frac { ( { \rm{e}}^{2i\omega t } + 2 ) ( \bar { \omega } + \bar{b } ) + { \rm{e}}^{2i\omega t } \frac{i}{2\bar{\tau}_{\rm{m } } } ) } { ( \bar { \omega } + \bar{b})^2 + ( \frac{1}{2\bar{\tau}_{\rm{m}}})^2 } \bigg ] \nonumber \\ & \times & \big[1- \frac{i}{\tilde s } \int \frac{d { { \mathbf{k}}}^{\prime}}{(2\pi ) ^3 } \int \frac{d{\omega}^{\prime } } { 2\pi } f_{\rm{b } } ( \omega^{\prime } ) \frac{-i/{\tau_{\rm{m}}}}{(\omega ^{\prime}-\omega _ { { \mathbf{k^{\prime}}}})^2+[1/(2\tau_{\rm{m}})]^2 } \big ] \nonumber \\ & \times & \bigg\ { \frac{1}{\big [ \bar{\omega } - \bar{\omega } -y^2- ( \bar{j}s+\frac{\bar{b}}{2}+\bar { \mu } ) \big]^2 + ( \frac{1}{2\bar{\tau}})^2 } \nonumber \\ & \times & \bigg [ \frac{i}{\bar{\tau } } \frac{1}{{\rm{e}}^{[{{y}}^2 + ( \bar{j}s+\bar{b}/2 ) -\bar{\mu } ] /{\bar{t}}}+1 } + \bar{\omega } - \bar{\omega } -y^2- ( \bar{j}s+\frac{\bar{b}}{2}+\bar { \mu } ) - \frac{i}{2\bar{\tau } } \bigg ] \nonumber \\ & \times & \frac{1}{\big [ \bar{\omega } -y^2 + ( \bar{j}s+\frac{\bar{b}}{2}+\bar { \mu } ) \big]^2 + ( \frac{1}{2\bar{\tau}})^2 } \bigg [ \frac{i}{\bar{\tau } } \frac{1}{{\rm{e}}^{[{{y}}^2 -(\bar{j}s+\bar{b}/2 ) -\bar{\mu } ] /{\bar{t}}}+1 } + \bar{\omega } -y^2 + ( \bar{j}s+\frac{\bar{b}}{2}+\bar { \mu } ) - \frac{i}{2\bar{\tau } } \bigg ] \nonumber \\ & - & \frac{1}{{\bar{\tau}}^2 } \frac{1}{\big [ \bar{\omega } - \bar{\omega } -y^2- ( \bar{j}s+\frac{\bar{b}}{2}+\bar { \mu } ) \big]^2 + ( \frac{1}{2\bar{\tau}})^2 } \cdot \frac{1}{{\rm{e}}^{[{{y}}^2 + ( \bar{j}s+\bar{b}/2 ) -\bar{\mu } ] /{\bar{t}}}+1 } \nonumber \\ & \times & \frac{1}{\big [ \bar{\omega } -y^2 + ( \bar{j}s+\frac{\bar{b}}{2}+\bar { \mu } ) \big]^2 + ( \frac{1}{2\bar{\tau}})^2 } \cdot \big ( 1- \frac{1}{{\rm{e}}^{[{{y}}^2 -(\bar{j}s+\bar{b}/2 ) -\bar{\mu } ] /{\bar{t}}}+1 } \big ) \bigg\ } \nonumber \\ % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % & + & { \bar{j } } \int d\bar { \omega } \int_0^{\infty } dy y^2 \bigg [ \frac { { \rm{e}}^{-2i\omega t } ( \bar { \omega } -\bar{b}+\frac{i}{2\bar{\tau}_{\rm{m } } } ) } { ( -\bar { \omega } -\bar{b})^2 + ( \frac{1}{2\bar{\tau}_{\rm{m}}})^2 } - \frac { ( { \rm{e}}^{-2i\omega t } + 2 ) ( -\bar { \omega } + \bar{b } ) + { \rm{e}}^{-2i\omega t } \frac{i}{2\bar{\tau}_{\rm{m } } } ) } { ( -\bar { \omega } + \bar{b})^2 + ( \frac{1}{2\bar{\tau}_{\rm{m}}})^2 } \bigg ] \nonumber \\ & \times & \big[1-\frac{i}{\tilde s } \int \frac{d { { \mathbf{k}}}^{\prime}}{(2\pi ) ^3 } \int \frac{d{\omega}^{\prime } } { 2\pi } f_{\rm{b } } ( \omega^{\prime } ) \frac{-i/{\tau_{\rm{m}}}}{(\omega ^{\prime}-\omega _ { { \mathbf{k^{\prime}}}})^2+[1/(2\tau_{\rm{m}})]^2 } \big ] \nonumber \\ & \times & \bigg\ { \frac{1}{\big [ \bar{\omega } + \bar{\omega } -y^2- ( \bar{j}s+\frac{\bar{b}}{2}+\bar { \mu } ) \big]^2 + ( \frac{1}{2\bar{\tau}})^2 } \nonumber \\ & \times & \bigg [ \frac{i}{\bar{\tau } } \frac{1}{{\rm{e}}^{[{{y}}^2 + ( \bar{j}s+\bar{b}/2 ) -\bar{\mu } ] /{\bar{t}}}+1 } + \bar{\omega } + \bar{\omega } -y^2- ( \bar{j}s+\frac{\bar{b}}{2}+\bar { \mu } ) - \frac{i}{2\bar{\tau } } \bigg ] \nonumber \\ & \times & \frac{1}{\big [ \bar{\omega } -y^2 + ( \bar{j}s+\frac{\bar{b}}{2}+\bar { \mu } ) \big]^2 + ( \frac{1}{2\bar{\tau}})^2 } \bigg [ \frac{i}{\bar{\tau } } \frac{1}{{\rm{e}}^{[{{y}}^2 -(\bar{j}s+\bar{b}/2 ) -\bar{\mu } ] /{\bar{t}}}+1 } + \bar{\omega } -y^2 + ( \bar{j}s+\frac{\bar{b}}{2}+\bar { \mu } ) - \frac{i}{2\bar{\tau } } \bigg ] \nonumber \\ & - & \frac{1}{{\bar{\tau}}^2 } \frac{1}{\big [ \bar{\omega } + \bar{\omega } -y^2- ( \bar{j}s+\frac{\bar{b}}{2}+\bar { \mu } ) \big]^2 + ( \frac{1}{2\bar{\tau}})^2 } \cdot \frac{1}{{\rm{e}}^{[{{y}}^2 + ( \bar{j}s+\bar{b}/2 ) -\bar{\mu } ] /{\bar{t}}}+1 } \nonumber \\ & \times & \frac{1}{\big [ \bar{\omega } -y^2 + ( \bar{j}s+\frac{\bar{b}}{2}+\bar { \mu } ) \big]^2 + ( \frac{1}{2\bar{\tau}})^2 } \cdot \big ( 1- \frac{1}{{\rm{e}}^{[{{y}}^2 -(\bar{j}s+\bar{b}/2 ) -\bar{\mu } ] /{\bar{t}}}+1 } \big ) \bigg\ } , \nonumber \\ \label{eqn : e100}\end{aligned}\ ] ] and @xmath222 a({\mathbf{x}},t ) c^{\dagger } ( { \mathbf{x}},t ) \sigma^{- } c({\mathbf{x}},t ) \bigg{\rangle } + \bigg{\langle } - \frac{\gamma ( t ) c^{\dagger } ( { \mathbf{x}},t ) \sigma^{- } c({\mathbf{x}},t ) } { 4i } \bigg{\rangle } \nonumber \\ % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % & = & - \frac { js } { 4 \pi } ( \frac{\gamma _ 0}{2})^2 \int \frac{d { { \mathbf{k}}}}{(2\pi ) ^3 } \int d\omega \big[1-\frac{i}{\tilde s } \int \frac{d { { \mathbf{k}}}^{\prime}}{(2\pi ) ^3 } \int \frac{d{\omega}^{\prime } } { 2\pi } { \rm{g}}^{\rm { < } } _ { { { \mathbf{k}}}^{\prime } , \omega ^{\prime } } \big ] \\ & \times & \bigg{\ { } % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % \big [ { \rm{e}}^{2i\omega t } { \rm{g}}^{\rm{r}}_{0 , \omega } + { \rm{g}}^{\rm{r}}_{0 , -\omega } + ( { \rm{e}}^{2i\omega t}+1 ) { \rm{g}}^{\rm{a}}_{0 , -\omega } \big ] \cdot \big[{\mathcal{g}}^{\rm{t}}_{\uparrow , { \mathbf{k } } , \omega -\omega } { \mathcal{g}}^{\rm{t}}_{\downarrow , { \mathbf{k } } , \omega } - { \mathcal{g}}^{\rm{<}}_{\uparrow , { \mathbf{k } } , \omega -\omega } { \mathcal{g}}^{\rm{>}}_{\downarrow , { \mathbf{k } } , \omega } \big ] \nonumber \\ % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % & + & \big [ { \rm{e}}^{-2i\omega t } { \rm{g}}^{\rm{r}}_{0 , -\omega } + { \rm{g}}^{\rm{r}}_{0 , \omega } + ( { \rm{e}}^{-2i\omega t}+1 ) { \rm{g}}^{\rm{a}}_{0 , \omega } \big ] \cdot \big[{\mathcal{g}}^{\rm{t}}_{\uparrow , { \mathbf{k } } , \omega + \omega } { \mathcal{g}}^{\rm{t}}_{\downarrow , { \mathbf{k } } , \omega } - { \mathcal{g}}^{\rm{<}}_{\uparrow , { \mathbf{k } } , \omega + \omega } { \mathcal{g}}^{\rm{>}}_{\downarrow , { \mathbf{k } } , \omega } \big ] \bigg{\ } } \nonumber \\ & + & { \cal{o}}(j^2 ) + { \cal{o}}({(\gamma_0)}^0 ) \\ % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % & = & - { \bar{j } } \int d\bar { \omega } \int_0^{\infty } dy y^2 \bigg [ \frac { { \rm{e}}^{2i\omega t } ( \bar { \omega } -\bar{b}-\frac{i}{2\bar{\tau}_{\rm{m } } } ) } { ( \bar { \omega } -\bar{b})^2 + ( \frac{1}{-2\bar{\tau}_{\rm{m}}})^2 } - \frac { ( { \rm{e}}^{2i\omega t } + 2 ) ( \bar { \omega } + \bar{b } ) - { \rm{e}}^{2i\omega t } \frac{i}{2\bar{\tau}_{\rm{m } } } ) } { ( \bar { \omega } + \bar{b})^2 + ( \frac{1}{-2\bar{\tau}_{\rm{m}}})^2 } \bigg ] \nonumber \\ & \times & \big[1-\frac{i}{\tilde s } \int \frac{d { { \mathbf{k}}}^{\prime}}{(2\pi ) ^3 } \int \frac{d{\omega}^{\prime } } { 2\pi } f_{\rm{b } } ( \omega^{\prime } ) \frac{-i/{\tau_{\rm{m}}}}{(\omega ^{\prime}-\omega _ { { \mathbf{k^{\prime}}}})^2+[1/(2\tau_{\rm{m}})]^2 } \big ] \nonumber \\ & \times & \bigg\ { \frac{1}{\big [ \bar{\omega } - \bar{\omega } -y^2 + ( \bar{j}s+\frac{\bar{b}}{2}+\bar { \mu } ) \big]^2 + ( \frac{1}{2\bar{\tau}})^2 } \nonumber \\ & \times & \bigg [ \frac{i}{\bar{\tau } } \frac{1}{{\rm{e}}^{[{{y}}^2 -(\bar{j}s+\bar{b}/2 ) -\bar{\mu } ] /{\bar{t}}}+1 } + \bar{\omega } - \bar{\omega } -y^2 + ( \bar{j}s+\frac{\bar{b}}{2}+\bar { \mu } ) - \frac{i}{2\bar{\tau } } \bigg ] \nonumber \\ & \times & \frac{1}{\big [ \bar{\omega } -y^2- ( \bar{j}s+\frac{\bar{b}}{2}+\bar { \mu } ) \big]^2 + ( \frac{1}{2\bar{\tau}})^2 } \bigg [ \frac{i}{\bar{\tau } } \frac{1}{{\rm{e}}^{[{{y}}^2 + ( \bar{j}s+\bar{b}/2 ) -\bar{\mu } ] /{\bar{t}}}+1 } + \bar{\omega } -y^2- ( \bar{j}s+\frac{\bar{b}}{2}+\bar { \mu } ) - \frac{i}{2\bar{\tau } } \bigg ] \nonumber \\ & - & \frac{1}{{\bar{\tau}}^2 } \frac{1}{\big [ \bar{\omega } - \bar{\omega } -y^2 + ( \bar{j}s+\frac{\bar{b}}{2}+\bar { \mu } ) \big]^2 + ( \frac{1}{2\bar{\tau}})^2 } \cdot \frac{1}{{\rm{e}}^{[{{y}}^2 -(\bar{j}s+\bar{b}/2 ) -\bar{\mu } ] /{\bar{t}}}+1 } \nonumber \\ & \times & \frac{1}{\big [ \bar{\omega } -y^2- ( \bar{j}s+\frac{\bar{b}}{2}+\bar { \mu } ) \big]^2 + ( \frac{1}{2\bar{\tau}})^2 } \cdot \big ( 1- \frac{1}{{\rm{e}}^{[{{y}}^2 + ( \bar{j}s+\bar{b}/2 ) -\bar{\mu } ] /{\bar{t}}}+1 }
\big ) \bigg\ } \nonumber \\ & - & { \bar{j } } \int d\bar { \omega } \int_0^{\infty } dy y^2 \bigg [ \frac { { \rm{e}}^{-2i\omega t } ( \bar { \omega } -\bar{b}-\frac{i}{2\bar{\tau}_{\rm{m } } } ) } { ( -\bar { \omega } -\bar{b})^2 + ( \frac{1}{-2\bar{\tau}_{\rm{m}}})^2 } - \frac { ( { \rm{e}}^{-2i\omega t } + 2 ) ( -\bar { \omega } + \bar{b } ) - { \rm{e}}^{-2i\omega t } \frac{i}{2\bar{\tau}_{\rm{m } } } ) } { ( -\bar { \omega } + \bar{b})^2 + ( \frac{1}{-2\bar{\tau}_{\rm{m}}})^2 } \bigg ] \nonumber \\ & \times & \big[1-\frac{i}{\tilde s } \int \frac{d { { \mathbf{k}}}^{\prime}}{(2\pi ) ^3 } \int \frac{d{\omega}^{\prime } } { 2\pi } f_{\rm{b } } ( \omega^{\prime } ) \frac{-i/{\tau_{\rm{m}}}}{(\omega ^{\prime}-\omega _ { { \mathbf{k^{\prime}}}})^2+[1/(2\tau_{\rm{m}})]^2 } \big ] \nonumber \\ & \times & \bigg\ { \frac{1}{\big [ \bar{\omega } + \bar{\omega } -y^2 + ( \bar{j}s+\frac{\bar{b}}{2}+\bar { \mu } ) \big]^2 + ( \frac{1}{2\bar{\tau}})^2 } \nonumber \\ & \times & \bigg [ \frac{i}{\bar{\tau } } \frac{1}{{\rm{e}}^{[{{y}}^2 -(\bar{j}s+\bar{b}/2 ) -\bar{\mu } ] /{\bar{t}}}+1 } + \bar{\omega } + \bar{\omega } -y^2 + ( \bar{j}s+\frac{\bar{b}}{2}+\bar { \mu } ) - \frac{i}{2\bar{\tau } } \bigg ] \nonumber \\ & \times & \frac{1}{\big [ \bar{\omega } -y^2- ( \bar{j}s+\frac{\bar{b}}{2}+\bar { \mu } ) \big]^2 + ( \frac{1}{2\bar{\tau}})^2 } \nonumber \\ & \times & \bigg [ \frac{i}{\bar{\tau } } \frac{1}{{\rm{e}}^{[{{y}}^2 + ( \bar{j}s+\bar{b}/2 ) -\bar{\mu } ] /{\bar{t}}}+1 } + \bar{\omega } -y^2- ( \bar{j}s+\frac{\bar{b}}{2}+\bar { \mu } ) - \frac{i}{2\bar{\tau } } \bigg ] \nonumber \\ & - & \frac{1}{{\bar{\tau}}^2 } \frac{1}{\big [ \bar{\omega } + \bar{\omega } -y^2 + ( \bar{j}s+\frac{\bar{b}}{2}+\bar { \mu } ) \big]^2 + ( \frac{1}{2\bar{\tau}})^2 } \cdot \frac{1}{{\rm{e}}^{[{{y}}^2 -(\bar{j}s+\bar{b}/2 ) -\bar{\mu } ] /{\bar{t}}}+1 } \nonumber \\ & \times & \frac{1}{\big [ \bar{\omega } -y^2- ( \bar{j}s+\frac{\bar{b}}{2}+\bar { \mu } ) \big]^2 + ( \frac{1}{2\bar{\tau}})^2 } \cdot \big ( 1- \frac{1}{{\rm{e}}^{[{{y}}^2 + ( \bar{j}s+\bar{b}/2 ) -\bar{\mu } ] /{\bar{t}}}+1 } \big ) \bigg\}. \label{eqn : e200 } \end{aligned}\ ] ] 10 i. zutic , j. fabian , and s. d. sarma : rev .
@xmath223 ( 2004 ) 323 .
y. tserkovnyak , a. brataas , g. e. w. bauer , and b. i. halperin : rev .
@xmath224 ( 2005 ) 1375 , and references therein .
y. kajiwara , k. harii , s. takahashi , j. ohe , k. uchida , m. mizuguchi , h. umezawa , h. kawai , k. ando , k. takanashi , s. maekawa , and e. saitoh : nature @xmath225 ( 2010 ) 262 . c. w. sandweg , y. kajiwara , a. v. chumak , a. a. serga , v. i. vasyuchka , m. b. jungfleisch , e. saitoh , and b. hillebrands : phys
@xmath226 ( 2011 ) 216601 .
h. kurebayashi , o. dzyapko , v. e. demidov , d. fang , a. j. ferguson , and s. o. demokritov : nat .
@xmath227 ( 2011 ) 660 .
e. simanek and b. heinrich : phys .
b @xmath228 ( 2003 ) 144418 . k. tsutsui , a. takeuchi , g. tatara , and s. murakami : j. phys .
@xmath229 ( 2011 ) 084701 .
s. zhang and z. li : phys .
lett . @xmath230 ( 2004 ) 127204 .
j. shi , p. zhang , x. xiao , and q. niu : phys .
lett . @xmath231
( 2006 ) 076604 .
l. mista : phys .
letters @xmath232 ( 1967 ) 646 .
r. c. ralph and m. d. stiles : j. magn .
magn . mater .
@xmath233 ( 2008 ) 1190 .
k. nakata and g. tatara : j. phys .
@xmath229 ( 2011 ) 054602 .
g. tatara ( private communication ) .
k. ando , s. takahashi , j. ieda , h. kurebayashi , t. trypiniotis , c. h. w. barnes , s. maekawa , and e. saitoh : nat .
@xmath227 ( 2011 ) 655 . j. rammer and h. smith : rev . mod .
@xmath234 ( 1986 ) 323 .
a. kamenev : _ field theory of non - equilibrium systems _
( cambridge university press , 2011 , arxiv:0412296 ) p. 31 .
t. kita : prog .
@xmath235 ( 2010 ) 581 .
h. haug and a. p. jauho : _ quantum kinetics in transport and optics of semiconductors _
( springer new york , 2007 ) p. 66 .
g. tatara , h. kohno , and j. shibata : phys .
rep . @xmath236 ( 2008 ) 213 .
d. a. ryndyk , r. gutierrez , b. song , and g. cuniberti : _ energy transfer dynamics in biomaterial systems _
( springer - verlag , 2009 , arxiv:0805.0628 ) p. 213 .
j. rammer : _ quantum field theory of non - equilibrium states _
( cambridge university press , 2007 ) p. 93 .
h. adachi , j. ohe , s. takahashi , and s. maekawa : phys .
b @xmath237 ( 2011 ) 094410 .
a. takeuchi , k. hosono , and g. tatara : phys .
b @xmath238 ( 2010 ) 144405 .
j. xiao , g. e. w. bauer , k. uchida , e. saitoh , and s. maekawa : phys .
b @xmath238 ( 2010 ) 214418 , and references therein . c. kittel : _ introduction to solid state physics _
( wiley , 1963 ) p. 49 . k. nakata : mod .
b @xmath239 ( 2012 ) 1250093/arxiv:1204.2339 .
h. kurebayashi , o. dzyapko , v. e. demidov , d. fang , a. j. ferguson , and s. o. demokritov : nat .
@xmath227 ( 2011 ) 660 . s. zhang and z. li : phys . rev .
@xmath230 ( 2004 ) 127204 .
a. thiaville , y. nakatani , j. miltat , and y. suzuki : europhys .
@xmath240 ( 2005 ) 990 .
k. ando , y. kajiwara , s. takahashi , s. maekawa , k. takemoto , m. takatsu , and e. saitoh : phys .
b @xmath241 ( 2008 ) 014413
. h. y. inoue , k. harii , k. ando , k. sasage , and e. saitoh : j. appl .
phys . @xmath242 ( 2007 ) 083915 .
a. brataas , y. tserkovnyak , g. e. w. bauer , and p. j. kelly : arxiv:1108.0385 .
q. zhang , s. hikino , and s. yunoki : appl .
@xmath243 ( 2011 ) 172105 .
k. xia , p. j. kelly , g. e. w. bauer , a. brataas , and i. turek : phys . rev .
b @xmath244 ( 2002 ) 220401(r ) .
t. l. gilbert : ieee transactions on magnetics @xmath245 ( 2004 ) 3443 .
h. kohno , g. tatara , and j. shibata : j. phys .
@xmath246 ( 2006 ) 113706 .
a. brataas , y. tserkovnyak , g. e. w. bauer and b. i. halperin : phys .
b @xmath247 ( 2002 ) 060404(r ) . | we theoretically propose quantum spin pumping mediated by magnons , under a time - dependent transverse magnetic field , at the interface between a ferromagnetic insulator and a non - magnetic metal .
the generation of a spin current under a thermal equilibrium condition is discussed by calculating the spin relaxation torque , which breaks the spin conservation law for conduction electrons and operates the coherent magnon state .
localized spins lose spin angular momentum by emitting magnons and conduction electrons flip from down to up by absorbing the momentum .
the spin relaxation torque has a resonance structure as a function of the angular frequency of the applied transverse field .
this fact is useful to enhance the spin pumping effect induced by quantum fluctuations .
we also discuss the distinction between our quantum spin pumping theory and the one proposed by tserkovnyak et al . |
since the existence of dirac quasiparticles has been proved for graphene , one of the most intriguing issues of its physics is the possibility of their localization by whichever imperfection that can appear in the honeycomb lattice.@xcite moreover , it should not be overlooked that the effect of disorder on the transport of quasiparticles is sensitive to the nature of inhomogeneity , be it caused by short - range or long - range defects , neutral or charged impurities , adsorbed or interstitial atoms or molecules , vacancies , spatial distortions or structural irregularities , including ripples on the graphene sheet and other long - wave random modulations .
all these different types of imperfections necessitate corresponding dedicated studies , which can not be accomplished by resorting to a single impurity model .
since these imperfections are rather dissimilar both in their character and action on basic properties of graphene , respective theoretical models should be diverse as well .
early experiments on graphene - based devices , which were engineered like a commonplace field effect transistors , revealed that the sample conductivity never drops below a certain _ minimal _
value.@xcite this fact , indeed , considerably reduced audacious expectations that corresponding devices are capable to serve as next - generation electronic switches .
nevertheless ( or , maybe , exactly because of this ) , the minimal conductivity existence has produced quite a stir , and its origin has been relentlessly debated .
distinctive features of charge carriers , which were shown to obey the linear dispersion , constituted the core of this discussion .
uniqueness of the electron subsystem in graphene were pushed to its limits so much that former physics of semiconductors were sometimes categorically declared being utterly unsuitable for this uncommon material .
it has been speculated that massless , according to their dirac dispersion , charge carriers can not be localized by any degree of disorder caused by lattice imperfections or impurity centers . the presumed impossibility to localize dirac excitations
were directly linked to the minimal conductivity phenomenon by a simple reasoning .
since the product of the wave vector modulus and the mean free path is confined from below for propagating states according to the ioffe - regel criterion , the conventional drude expression immediately yields the minimum value of @xmath0 for each conducting channel .
this argument , along with available experimental results , has led to an opinion that the minimum conductivity value has a _ universal _ character for graphene , and is expressed only through fundamental constants . in the mean time , there appeared occasional theoretical studies , which did not deny a possibility to localize charge carriers in graphene , and argued for the mobility edge appearance under certain circumstances , either for a disorder of a general type,@xcite or for a specific impurity model , in particular , for substitutional defects,@xcite and for the anderson model of disorder.@xcite later , it has become apparent that the minimal conductivity in graphene does depend on the sample quality and noticeably varies with the impurity concentration.@xcite furthermore , not so long ago , a metal - insulator transition ( mit ) in graphene dosed with atomic , which is essential , hydrogen has been convincingly observed .
the mit has been manifested by an increase of the room temperature resistance in about four orders of magnitude.@xcite the transition has been reported for the grown on sic surface graphene with a low hydrogen coverage , namely , around @xmath1 .
the presence of the mobility gap has been also reported for the fluorinated and ion bombarded graphene.@xcite as well , corresponding data of arpes measurements indicated the disappearance of quasiparticle excitations in the system .
the mit of the anderson type comes about when the fermi level finds itself inside a domain of _ localized _ states . since the graphene in the actual experiment was pristine before the hydrogen deposition and possessed sounding metallic properties with conventional fermi - liquid behavior , it is tempting to anticipate that some amount of deposited hydrogen is sufficient to open up a reasonably wide region filled with localized states , or a _ quasigap _ , in the spectrum .
in essence , this draft description of the leading to the mit process corresponds to the well - known phenomenon of _ spectrum rearrangement_,@xcite which deals with decisive modifications in the elementary excitation spectrum upon increasing the impurity concentration in the system .
when the quasigap is located in the vicinity of the fermi level position , the mit can take place in spite of the rather low hydrogen concentration .
below we are going to link together the predicted spectrum rearrangement in graphene with point defects and the observed mit under the hydrogen dosing.@xcite the spectrum rearrangement takes place when a single impurity induces a local or a resonance level in the spectrum .
it has been demonstrated recently that graphene almost inevitably contains traces of impurity centers , which are capable in producing resonance states.@xcite
let us discuss some general spectral properties of impure systems and assume for this purpose that inside an isolated host band in a hypothetical @xmath2-dimensional system a single impurity center accounts for a resonance state with the energy @xmath3 , which is measured against the band edge and has been made dimensionless by the corresponding bandwidth .
the dispersion relation for this host band is supposed to be of the form @xmath4 . then , the spatial behavior of the host green s function at a given energy inside the band , apart from any other details , should be governed by oscillations with the characteristic radius @xmath5 , which is expressed in units of the appropriate lattice constant . at the energy @xmath6
, these oscillations determine the effective radius @xmath7 of the impurity state . through a closeness of the resonance energy to the band edge
, the effective radius of the impurity state may far exceed the lattice period .
as the impurity concentration is increased , the individual impurity states are becoming more tightly packed .
it can be anticipated that a significant spatial overlap between these states is capable in provoking qualitative changes in the spectrum .
the average distance between impurities @xmath8 , where @xmath9 is the ( relative ) impurity concentration , is gradually shrinking with increasing @xmath9 , and at some stage of the process becomes of the order of @xmath10 .
this general condition defines the critical concentration @xmath11 of the spectrum rearrangement in the corresponding impure system .
it seems justifiable to reiterate here that the value of @xmath12 can be fairly low due to the smallness of the energy @xmath13 .
graphene is , evidently , a two - dimensional system , so that @xmath14 , and features the linear dispersion of charge carriers , which implies that @xmath15 .
consequently , the critical concentration of spectrum rearrangement in the impure graphene is expected to be @xmath16 , where energy is counted from the dirac point , at which the valence band and the conduction band coincide . in a case of the conventional dispersion with @xmath17 ,
the identical in appearance relation between @xmath6 and @xmath12 is inherent in four - dimensional systems . in this sense graphene ,
as regards the spectrum rearrangement , can be formally viewed as a conventional system of _ increased _ spatial dimensionality .
this fact further accentuates the uniqueness of graphene , since the spectrum rearrangement has not been studied for such systems .
let us return back to experimental data contained in ref . .
the comparison of angle - integrated spectrum for clean and hydrogen dosed graphene suggests that the dimensional resonance energy @xmath18 lies somewhere around @xmath19 ev lover than the initial fermi level , while the dirac point position @xmath20 in the undoped crystal is about @xmath21 ev below it .
a bit more careful analysis of the spectrum rearrangement in impure graphene , @xcite which is based on the expansion of the self - energy into a cluster series,@xcite brings about the following expression for @xmath12 , which , for a convenience , is rewritten here in terms of the hydrogen coverage , @xmath22 where @xmath23 is the areal density of carbon atoms in graphene , @xmath24 nm the lattice constant of graphene , @xmath25 the critical coverage of hydrogen atoms , @xmath26 ev the bandwidth parameter , and the logarithmic correction is omitted because of the relatively large resonance energy .
substitution of the guessed value for the resonance energy results in @xmath25 that is reasonably close to the reported mit critical hydrogen coverage of @xmath27 @xmath28.@xcite as a matter of fact , the critical coverage of the spectrum rearrangement @xmath25 , should not be identical to the experimentally obtained hydrogen coverage required for the mit , which will be discussed later .
it should be also noted that there are two main types of the spectrum rearrangement : the _ cross _ one , which is usually accompanied by a sharp single impurity resonance , and the _ anomalous _ one . the first type of the spectrum rearrangement results in a quasiparticle dispersion that looks like a familiar hybridization between the host branch and a dispersionless branch at the impurity resonance energy . in this case ,
two new branches are separated by a gap , which widens with an increase in the impurity concentration .
consequently , two different wave vectors correspond to the same energy in this split spectrum .
however , this feature has not been detected in the arpes measurements on the hydrogenated graphene .
the second type of the spectrum rearrangement , which is usually encountered in low - dimensional systems , is of a more diffused nature and frequently corresponds to a considerably smeared single impurity resonance .
this anomalous spectrum rearrangement is characterized by the opening of the quasigap , which is filled with localized states and separates two non - overlapping branches of extended states exhibiting a renormalized dispersion .
the anomalous type of the spectrum rearrangement has been shown to unfold in graphene for impurities described by the lifshitz model.@xcite within this model , identical impurity centers are supposed to be randomly distributed on sites of the host lattice .
each impurity is only allowed to change the on - site energy at its location in the corresponding tight - binding hamiltonian .
it is well - known that the electron subsystem in graphene encompasses practically free @xmath29-electrons , which number is equal to the number of lattice sites .
as soon as an additional hydrogen atom lands on graphene , its uncoupled s - electron immediately enters a chemical bond with one of the @xmath29-electrons .
that should make the latter one localized almost completely at the impurity site . in the first approximation
, one can assume that the resulting perturbation comes only from the appearance of strong attracting potential on the occupied carbon atom .
thus , the disordered system is described by following operators , @xmath30 where @xmath31 is the host hamiltonian , in which summation is restricted to nearest neighbors , @xmath32 runs over lattice cells , indices @xmath33 and @xmath34 enumerate sublattices , @xmath35 and @xmath36 are the creation and annihilation operators at the corresponding lattice site , @xmath37 ev is the tight - binding transfer integral for the @xmath29 bands in graphene , the variable @xmath38 takes the value of @xmath39 with the probability @xmath40 or the value of @xmath41 with the probability @xmath42 , where @xmath43 is the hydrogen coverage , and @xmath44 is the difference in the on - site potential at the defect position . within this model ,
the local density of states ( ldos ) at the lattice site occupied by a hydrogen reads , @xmath45 where @xmath46 is the diagonal element of the host green s function @xmath47 in the vicinity of the dirac point in the spectrum , namely , within the window stretching up to around @xmath48 ev to each side of it , this diagonal element can be easily approximated , @xmath49 where @xmath50 is the same bandwidth parameter as in ( [ critcov ] ) .
substitution of this approximation to eq .
( [ ldos ] ) yields @xmath51 } , \label{ldosappr}\ ] ] where @xmath52 the ldos at the impurity site is shown in fig .
[ f1 ] for @xmath53 ev . as follows from the figure , the resonance energy @xmath18 is located approximately @xmath54 ev above the dirac point and thus corresponds to the experimentally observed peak . while the required change in the on - site potential is quite substantial ,
the large perturbation magnitude should be considered , first of all , as a result of the chosen simplified impurity model .
unless another is explicitly specified , the perturbation value of @xmath55 ev will be used below for all subsequent estimations .
ev.,scaledwidth=96.0% ] when a certain amount of deposited hydrogen atoms is taken into account , the negative potentials induced by them on the occupied lattice sites impede the electron movement . for the lifshitz impurity model , which has been described above ,
the course of the spectrum rearrangement has been studied both analytically and numerically.@xcite the analysis has shown that states are subject to localization near the resonance energy , where the impurity scattering is the strongest . at some coverage ,
a quasigap occupied by localized states opens around the energy @xmath18 . with increasing impurity concentration
, the quasigap gradually broadens and the degree of localization inside it rises .
this quasigap expansion and the localization enhancement are not symmetric about the energy @xmath18 and are more expressed above the resonance ( for an attractive potential ) .
finally , the quasigap consumes all the space between @xmath18 and @xmath20 in the spectrum , which corresponds to the onset of the spectrum rearrangement by the definition . in other words ,
the spectrum rearrangement does follow the anomalous scenario . in the experiment discussed
, dosed hydrogen atoms act not only as scattering centers , but also as acceptors .
thus , their presence inevitably lowers the fermi level of the system .
therefore , while the hydrogen coverage is increasing , the fermi level and the upper boundary of the quasigap , where the localization of states is most pronounced , are moving toward each other . sooner or later , the fermi level should appear inside the quasigap , which will cause the mit of the anderson type . since the energy @xmath18 is positioned somewhere in - between @xmath20 and the bare fermi level @xmath56 , the critical coverage of the spectrum rearrangement should be close but not identical to the critical coverage of the mit .
the acceptor effect of the deposited hydrogen atoms is also implicitly contained in the lifshitz model .
strong attracting potential @xmath44 produces a deep local level below the conduction band .
because of its remoteness from the continuous spectrum , the corresponding narrow impurity band should hold almost @xmath57 energy levels , where @xmath58 is the total amount of carbon atoms in the system .
thus , taking into account the spin degeneracy , this impurity band consumes nearly one electron per deposited hydrogen atom .
however , apart from these qualitative considerations , the adopted impurity model allows to address the question on the fermi level position more precisely .
consider the ( normalized ) total number of states with energies that are less than a specified one , @xmath59 where @xmath60 is the density of states . in a disordered system
this quantity can be expanded in powers of the impurity concentration,@xcite @xmath61 where the presence of two sublattices in graphene and the lifshitz impurity model are taken into account , and @xmath62 is the total number of states in the host system . keeping in mind that the deep impurity band is modeling the acceptor effect of impurities , it is natural to demand the conservation of the number of occupied states in the system with varying the amount of dosed hydrogen .
this yields a kind of a balance condition , @xmath63 where the constant term is dropped from the both sides of the equation , and eq . ( [ apr ] ) is used to obtain @xmath62 for the clean graphene . from here and on this condition
will be employed to determine the fermi level position at a given dosing level .
it is worth mentioning that the dirac point position ( which should not be confused with its value for the unperturbed system @xmath20 ) is gradually shifted with increasing the impurity concentration by approximately @xmath64 . since the dirac point and the fermi level
are both moving in one direction , namely to lower energies , the distance between them does not shorten remarkably .
for @xmath65 , the presence of attracting impurities does not distort significantly the quasiparticle dispersion in the valence band .
in contrast , impurities noticeably affect the conduction band .
the self - energy @xmath66 , which enters the dyson equation @xmath67 for the averaged over impurity distributions green s function of the disordered system @xmath68 can not be attributed only to the impurity scattering in the actual experiment of ref . , because the linewidth of the arpes data is not negligible even at the absence of impurities . to deal with this issue
, we assume that all other sources of its broadening , except the controlled amount of hydrogen atoms , can be roughly described by a constant dumping term , @xmath69 where @xmath70 is those part of the self - energy , which results from the impurity scattering , and @xmath71 is the identity matrix .
a comparison of the density of states obtained by the numerical simulation with the one resulted from the standard average t - matrix method showed that at @xmath65 this approximation works rather well in the vicinity of the impurity resonance.@xcite within this single - site approximation , the impurity - induced part of the self - energy is diagonal in lattice sites and sublattices for point defects , @xmath72 thus , taking into account both assumptions , the total self - energy is also diagonal , @xmath73 according to the conventional expression for the spectral function , @xmath74 where @xmath75 is the fermi velocity , the wave vector @xmath76 is counted from the corresponding dirac point , and only one branch , which is manifested in the actual experiment , is retained , it is straightforward to put down corresponding expressions for the inverse width of the momentum distribution curve at its half - height , @xmath77 and for the inverse of its median @xmath78 both this quantities are taken at the fermi level , @xmath79 and @xmath80 , and depicted in fig .
[ f2 ] in their dependence on the hydrogen coverage .
the fermi level position is calculated from eq .
( [ ef ] ) , and the external damping @xmath81 is chosen to be @xmath82 ev .
( solid line ) and the inverse wave vector @xmath83 ( dashed line ) at the fermi level _ vs _ the hydrogen coverage.,scaledwidth=96.0% ] the arpes linewidth analysis given in fig .
3(c ) of ref .
can be compared with fig .
since the inverse momentum width @xmath84 corresponds to the magnitude of the mean free path of charge carriers , the ioffe - regel criterion@xcite is violated for states at the fermi level , _
i.e. _ @xmath85 , at somewhat higher hydrogen coverage , namely , around @xmath86 @xmath28 , than in the experiment
@xmath87 @xmath28 .
this discrepancy , for the most part , arise from the noticeably reduced wave vectors in the conduction band of the sample containing no hydrogen .
the reduction is evident in comparison with the dispersion in the valence band , or with the idealized graphene ( [ ham0 ] ) .
the experimentally obtained quasiparticle dispersion looks already distinctly distorted before any hydrogen deposition .
the distortion can be attributed to the possible presence of unidentified impurities in the system.@xcite to a certain extent , this issue can be crudely addressed by a corresponding adjustment of the hopping integral @xmath88 in the hamiltonian .
results of such quick guesstimate are shown in fig .
[ f3 ] .
( solid line ) and the inverse wave vector @xmath83 ( dashed line ) at the fermi level _ vs _ the hydrogen coverage for @xmath89 ev , @xmath90 ev , and @xmath91 ev.,scaledwidth=96.0% ] the deliberate adjustment of the transfer integral @xmath88 pushes @xmath83 to where it should be at the zero hydrogen coverage .
correspondingly , the ioffe - regel criterion is violated at lower hydrogen coverage matching the experimental data .
however , such a forthright measure will be suitable only in the fermi level vicinity .
it is more consistent to work out a proper model of the host system that is capable in simulating its experimentally obtained spectrum .
anyhow , the resulting adjustment of the zero coverage fermi wave vector moves the entire curve for the inverse fermi vector upwards and , thus , diminish the impurity concentration , at which the ioffe - regel criterion ceases to hold .
it should be mentioned that , according to the routine of the spectrum rearrangement analysis , one should expect the fermi level to enter inside a domain of localized states at @xmath92 .
this will happen for the idealized host spectrum at the hydrogen coverage around @xmath93 @xmath28 , which is a little bit higher than the coverage corresponding to @xmath85 . with further increase in the impurity concentration ,
the average t - matrix approximation ( [ tm ] ) becomes non - valid because of the increased scattering on impurity clusters , and the approach outlined above is only suitable for signaling a strong localization at the fermi level .
on the other side , the experimentally detected sharp increase in the sample resistance , see fig .
3(b ) of the ref .
, also succeeds the ioffe - regel criterion violation . within the kubo approach the zero temperature conductivity can be written as follows,@xcite @xmath94 , \label{kub}\end{gathered}\ ] ] where @xmath95 the dependence of the dimensionless conductivity @xmath96 on the hydrogen coverage
is plotted in fig .
[ f4 ] .
it is evident from fig .
[ f4 ] that the conductivity is falling down nearly exponentially with increasing the impurity concentration .
indeed , the kubo formula and the average t - matrix approximation are becoming not so reliable as approaching the mobility edge .
thus , the zero temperature conductivity can not be satisfactorily described by eqs .
( [ tm ] ) and ( [ kub ] ) close to the mit .
the concentration dependence of the energy distribution curve at the fermi wave vector @xmath97 can be easily re - created for the lifshitz impurity model with the help of eqs .
( [ ef ] ) , ( [ sg ] ) , ( [ tm ] ) , and ( [ sf ] ) . the corresponding magnitude , @xmath98 , is given in fig .
[ f5 ] as a function of the binding energy @xmath99 and the hydrogen coverage @xmath43 . with increasing the hydrogen coverage.,scaledwidth=96.0% ] it is clearly seen from the figure that the energy width of the quasiparticle peak at @xmath97 is gradually increasing with increasing the hydrogen coverage . close to the critical impurity concentration for the mit , the width of the energy distribution curve occupies nearly all conduction band below the fermi level , which indicates the presence of a developed quasigap in the spectrum .
this broadening translates into the complete breakdown of the quasiparticle picture near the mit .
the characteristic curve shape consisting of two competing peaks reproduces well the experimentally observable one , which can be seen in fig .
2(c ) of ref . .
at that , the peak , which is developing with increasing the hydrogen coverage , is connected with the impurity resonance .
finally , we render the density plots of the spectral function ( [ sf ] ) at hydrogen coverages that are close to the critical one for the mit in order to reproduce fig .
1(d ) of ref . .
[ f6 ] the fermi level is about to enter the spectral region near the resonance energy , where the quasigap is forming due to strong impurity scattering . in fig .
[ f7 ] the fermi level is already inside the developed quasigap . the pattern of fig .
[ f7 ] suggests the apparent upturn of the dispersion as approaching the fermi level of the system , which has been noticed in the experiment .
it is worth mentioning here that point defects does not cause a uniform broadening of the spectral function , on the contrary , the impurity - induced broadening is most pronounced in the vicinity of the resonance energy @xmath18 in this case .
@xmath28.,scaledwidth=96.0% ] @xmath28.,scaledwidth=96.0% ]
to summarize , we have established that point defects are introducing a new length parameter into the impure graphene .
this length parameter results from the presence of a single - impurity resonance and shows up as the effective radius of a single - impurity state .
when the average distance between impurities decreases up to this effective radius , the quasiparticle spectrum undergoes the cardinal rearrangement .
the spectrum rearrangement is manifested in the opening of the quasigap around the impurity resonance energy in the spectrum .
the quasigap progressively widens with increasing the impurity concentration .
the fermi level moves due to the doping effect of impurities , as well as confining the quasigap mobility edges are moving due to its expansion .
if the fermi level crosses one of the mobility edges and enters the quasigap , where states are localized , the mit of the anderson type takes place in the disordered system .
we have presented arguments , which confirm that the above scenario can be considered as a sounding candidate for the explanation of the experimentally observed mit in graphen dosed with hydrogen . even employing the simple lifshitz model for the impurity centers
, it appears possible to achieve a semi - quantitative interpretation of the experimental data .
the mit of the anderson type in this case is prompted by the gradual lowering of the fermi level due to the acceptor effect of the hydrogen atoms combined with the persistent raising of the mobility edge due to the ongoing spectrum rearrangement .
thus , the spectrum rearrangement acts as the main cause of the mit , and , as a common phenomenon in semiconductors , provides the basis for understanding the physics of the process .
indeed , the oversimplified impurity model is not able to convey all the detail of the system s behavior . to fulfill this task ,
more sophisticated impurity models are required . among them
the two - parametric s - d model looks like a more natural choice for the deposited hydrogen atom .
however , more sophisticated impurity models should not change the general physics of the transition process , which is already captured by the lifshitz impurity model .
this work was supported by the scopes grant no .
iz73z0 - 128026 of swiss nsf , the simtech grant no .
246937 of the european fp7 program , the state program `` nanotechnologies and nanomaterials '' , project no . 1.1.1.3 , and by the program of fundamental research of the department of physics and astronomy of the national academy of sciences of ukraine . | we demonstrate that the spectrum rearrangement can be considered as a precursor and a basis for the metal - insulator transition observed in graphene dosed with hydrogen atoms .
the anderson - type transition is attributed to the fermi level s entering into the quasigap , which develop in the vicinity of the impurity resonance energy due to the anomalous spectrum rearrangement .
theoretical results for the lifshitz impurity model are in a reasonable agreement with available experimental data . |
in mesoscopic samples of disordered normal metals at low temperatures it is possible to observe a quantum correction to the classical boltzmann conductance @xcite . here mesoscopic means that within the sample , the interaction of a single electron with other degrees of freedom , such as other electrons , phonons , magnetic impurities etc .
, can be neglected .
such a situation can be realised in semiconductor and metal nanostructures cooled to milli - kelvin temperatures .
the origin of the quantum correction is quantum interference between time reversed scattering paths .
( by `` time reversed path ''
we mean that the electron follows the same trajectory but in the opposite sense . )
the correction is sensitive to the breaking of spin rotation symmetry and is suppressed by the breaking of time reversal symmetry @xcite .
the spin orbit interaction has the important property of breaking spin rotation symmetry while preserving time reversal symmetry . as an example , the two probe conductance @xmath0 of a quasi - one dimensional mesoscopic wire can be expressed in the form @xmath1 @xcite .
the classical conductance of the wire is @xmath2 : @xmath3 where @xmath4 is the number of scattering channels and @xmath5 is the ratio of the length @xmath6 of the wire to the elastic mean free path @xmath7 . in the absence of an applied magnetic field the quantum correction @xmath8 , also called the `` weak localisation correction '' , is @xmath9 .
more exactly , in the diffusive regime @xmath10 we have @xmath11 if the spin orbit interaction is negligible and @xmath12 if it is not . in a magnetic field
sufficient to break time reversal symmetry the quantum correction is suppressed and becomes @xmath13 .
more recently , a quantum correction to the classical conductance of a disordered mesoscopic normal superconducting ( ns ) junction was observed @xcite .
the correction is most pronounced in junctions where the transparency @xmath14 of the schottky barrier at the ns interface is low @xmath15 and the length of the normal part is such that @xmath16 . under these conditions a phenomenon known as `` reflectionless tunnelling '' occurs . in @xcite
it is suggested that quantum interference is responsible for this effect .
the type of scattering paths involved in this quantum interference contain a segment where the path of an electron ( hole ) incident on the ns boundary is subsequently retraced by an andreev reflected hole ( electron . )
the enhancement of the junction conductance can be as much as several times greater than the classical conductance . in this paper
we investigate the effect of the coupling of the spin and spatial degrees of freedom of the electrons by an appreciable spin orbit interaction has on this quantum correction .
as previously mentioned , this has an important effect on the quantum correction in a normal metal .
as the electron diffuses through the metal , the spin orbit coupling causes a simultaneous diffusion of the direction of its spin .
interference between different spin states is now possible and this modifies the interference responsible for the weak localisation correction . what role then , if any , does this spin diffusion play in the ns junction ? at first sight the answer to this question would be appear to be none at all .
the argument is as follows : the spin rotations of an electron and an andreev reflected hole , which traverse time reversed paths , are exactly opposite and cancel each other . since interference between scattering processes involving such paths is believed to be responsible for the quantum correction , it should be insensitive to the spin orbit interaction . as we shall demonstrate below this
is too simple and in fact the spin orbit interaction does affect the quantum correction to the classical conductance of an ns junction .
we have found the flaw in the argument is that it is not only processes involving electrons and holes traversing time reversed paths which are responsible for the quantum correction . processes in which electrons and holes traverse loops in the same sense , what we shall call here identical paths , also contribute and in this case the spin rotations do not cancel each other .
the most striking consequence of the existence of interference involving such paths is that the quantum correction is sensitive to the spin orbit interaction even in a magnetic field . in the normal metal
we have seen that the quantum correction to the classical conductance is suppressed when time reversal symmetry is broken in an applied magnetic field .
this is easily understood since the symmetry between time reversed paths is broken when the time reversal symmetry is broken . in the ns junction however , the contribution of identical paths is still present even in a magnetic field ; the electron and hole carry opposite charges , and so the aharonov - bohm phases that they accumulate as they follow such trajectories cancel . moreover , since the electron and hole undergo identical spin rotations , as they follow identical paths , this contribution is sensitive to the spin diffusion induced by the spin orbit interaction .
the paper is organised as follows . in sections
[ bdg ] -[formulae ] we deal with some necessary preliminaries : the bogolubov - de gennes equations , scattering theory and conductance formulae for the ns junction , generalising the standard theory to take into account the spin orbit interaction . in section [ quasiball ]
we investigate a quasi - ballistic ns junction , that is to say a junction whose length is shorter than the mean free path but long enough so that its scattering matrix is well described by a certain random phase approximation @xcite .
we express the junction conductance @xmath17 as a sum of two terms @xmath18 a classical conductance @xmath19 and a quantum correction @xmath20 due to quantum interference .
we determine the quantum correction for the four ensembles listed in table [ table1 ] and the results are presented in table [ table2 ] . in section [ interp ]
we present a semiclassical interpretation of these results which permits us to identify the type of scattering paths which interfere to produce the quantum correction . in agreement with ref @xcite
, we find that there is an important contribution due to interference between processes involving electrons and holes traversing time reversed paths , but that there is also a second contribution from processes involving electrons and holes traversing identical paths .
we explain why this contribution is sensitive to the breaking of spin rotation symmetry but not to the breaking of time reversal symmetry . after reaching a firm understanding of the quasi - ballistic junction we consider the diffusive regime , looking first at junctions without a schottky barrier at the ns interface in section [ diffjunc ] . in zero magnetic field
the quantum correction is known to be of @xmath9 @xcite . in a magnetic field
the authors of ref @xcite find that the correction though smaller is still @xmath9 .
the effect of the spin orbit interaction was not considered in ref @xcite and so we repeat their calculation taking it into account .
we find that breaking spin rotation symmetry multiplies the correction by a factor of minus one half , regardless of whether time reversal symmetry is broken or not . in section [ reftun ]
we discuss the dramatic enhancement of the junction conductance , known as `` reflectionless tunnelling , '' which is observable when @xmath15 and @xmath16 . under these conditions
the contribution of time reversed paths is @xmath2 and dominates that of identical paths which is @xmath9 so that the reflectionless tunnelling effect is , in a first approximation , insensitive to the spin orbit interaction .
this conclusion has been confirmed by carrying out a numerical simulation of a junction under the relevant conditions .
the bogolubov - de gennes ( bdg ) equations @xcite appropriate for a metal where the electrons spatial and spin degrees of freedom are coupled by a significant spin - orbit interaction , are @xmath21 \left [ \begin{array}{c } \psi^{e } \\ \psi^{h } \end{array } \right ] = \epsilon \left [ \begin{array}{c } \psi^{e } \\ \psi^{h } \end{array } \right]\ ] ] where @xmath22 here @xmath23 is the single electron hamiltonian of the metal incorporating the spin - orbit interaction and @xmath24 is the fermi energy . in deriving this equation , it is assumed that an attractive point like interaction exists between the electrons : @xmath25 the full interacting hamiltonian is reduced to an effective non - interacting hamiltonian by introducing effective potentials @xmath26 @xmath27 where the overline indicates a thermal average with respect to the fermi distribution function and @xmath28 is the usual field operator appearing in the second quantised formulation of the interacting electron problem .
the time reversal operator @xmath29 , which appears in the bdg equations has the form @xmath30 @xmath31 with @xmath32 the operation of complex conjugation .
the eigenstates of the bdg equation describe the excitations of the interacting electron system .
the meaning of the electron @xmath33 and @xmath34 wavefunctions can be seen by writing the field operator as @xmath35 \gamma_n^{\dag } \right]\ ] ] where @xmath36 is the annihilation operator of the excitation labelled @xmath37 .
the corresponding eigenvalue @xmath38 of the bdg equations corresponds to the energy of the excitation . with the aid of ( [ bog.val.transf . ] ) the occupation of the eigenstates of @xmath39 for a given excitation can be determined .
in this section we develop the scattering theory appropriate to the normal superconducting junction shown schematically in figure 1 .
we treat the scattering at the ns interface within andreev s approximation .
we suppose that any impurities in the system are in the region indicated by shading in figure 1 , and we make the approximation that the magnetic field is zero everywhere except in this disordered region .
this is reasonable for the low fields of interest which affect the interference between electrons and holes in the normal part of the junction . to facilitate the explanation of the formalism ,
it is helpful to develop the scattering theory for a definite model .
we have chosen a lattice tight binding model , the same model which we will use later in numerical simulations .
an exact analogous explanation is also possible for a continuum model .
we consider a cubic lattice and take into account nearest neighbour interactions only .
we denote by @xmath40 the amplitude that the electron is in an s- orbital at @xmath41 with spin @xmath42 and similarly @xmath43 for a hole .
we include in the hamiltonian a spin orbit term which arises from the zeeman coupling of the electron spin with the effective magnetic field felt by the electron as it moves in the spatially varying potential of the lattice .
we ignore the direct zeeman coupling of the spin to any external magnetic field .
a simple calculation shows that the hamiltonian has the form : @xmath44 where @xmath45_{\sigma,\mu } \\
v^y_{\sigma,\mu } & = v_0 \delta_{\sigma,\mu } - v_1 i [ \sigma_y]_{\sigma,\mu } \\
v^z_{\sigma,\mu } & = v_0 \delta_{\sigma,\mu } - v_1 i [ \sigma_z]_{\sigma,\mu } \end{array}\ ] ] an external magnetic field @xmath46 , applied in the @xmath47 direction is modelled by peierl s factors in the matrix elements between nearest neighbours in the @xmath48 direction . if @xmath49 is lattice constant @xmath50 where @xmath51 is the flux quantum .
the transverse dimensions are @xmath52 and @xmath53 .
the hamiltonian ( [ model ] ) can be regarded as a three dimensional generalisation of that proposed in ref .
@xcite as a model for a two dimensional electron gas formed at the surface of a iii - v semiconductor .
the relevant energy scale of the model is determined by @xmath54 . for convenience we shall set this to unity with the choice @xmath55 varying the angle @xmath56
, we may set an arbitrary ratio of normal potential coupling @xmath57 to spin orbit coupling @xmath58 , while keeping the extent in energy of the density of states roughly constant . with this choice ,
@xmath59,@xmath60 and @xmath61 are all elements of @xmath62 . using the homomorphism of @xmath62 with the three dimensional rotation group @xmath63
, we can interpret the @xmath64 s as rotations of the spin of the electron as it moves between nearest neighbours @xcite .
a product of nearest neighbour matrix elements along a path will have the form @xmath65 where @xmath66 is the aharonov bohm phase picked up by the electron , and @xmath67 is the rotation of the electron s spin as it traverses the path .
the hamiltonian @xmath39 has the form given in ( [ model ] ) everywhere except in the disordered region located in @xmath68 .
there , some or all of the hamiltonian matrix elements are supposed random . in ref .
@xcite the diagonal elements were assumed to be independently and identically distributed , while the parameter @xmath56 controlling the spin orbit interaction was held fixed . in refs .
@xcite a random spin orbit interaction was also considered .
for the present purpose we do not need to specify the precise distribution .
first , we consider the scattering of electrons incident at an energy @xmath69 . to the left of the disordered section we expand the electron wavefunction @xmath70 in terms of the bloch states of ( [ model ] ) with energy @xmath71 .
@xmath72 @xmath73 as @xmath74 , far from the disordered region , we impose the boundary condition that the allowed states consist exclusively of incoming and outgoing propagating waves . thus in @xmath75 states with @xmath76
are excluded .
we denote by @xmath77 the number of open channels " , i.e. states with @xmath78 that carry a positive probability current in the @xmath79 direction ; there are an equal number carrying current in the @xmath80 direction .
we label these states so that @xmath81 carries a current in the @xmath79 direction and @xmath82 a current in the @xmath80 direction . after a suitable normalisation of the transverse wavefunctions ( see appendix [ a ] ) the electric current due to electrons at the left of the disordered section is @xmath83 the boundary condition as @xmath84 , imposed above , ensures that only open channels , and not `` closed channels '' with @xmath85 , contribute to the current .
a similar expansion may be made on the right in terms of a set of coefficients @xmath86 with the boundary condition that we admit only those states with @xmath87 .
thus , far to the right of the disordered section , as @xmath88 , the allowed states again consist exclusively of incoming and outgoing propagating waves . the @xmath89 scattering matrix for electrons
@xmath90 relates the @xmath91 incoming flux amplitudes at the left , @xmath92 and the right @xmath93 with the @xmath91 outgoing flux amplitudes at the left @xmath94 and the right @xmath95 @xmath96 = \left [ \begin{array}{c } a_- \\ a_+^{\prime } \end{array } \right]\ ] ] the matrix @xmath90 has the structure @xmath97\ ] ] in terms of the @xmath98reflection and transmission matrices for left incidence ( @xmath99 ) and right incidence ( @xmath100 ) .
since we are considering time independent scattering , the currents to the left and right of the disordered section must be equal , and therefore it follows that @xmath90 is unitary .
there is an additional restriction on @xmath90 when , in the absence of an applied magnetic field , the hamiltonian is time reversal invariant ie .
@xmath101 = 0 $ ] with @xmath29 given in eq .
( [ t ] ) . for a suitable choice of transverse wavefunctions ( see appendix [ a ] ) @xmath102 will then satisfy @xmath103 s_e \left[\begin{array}{cc } -1_{2n } & 0 \\ 0 & 1_{2n }
\end{array}\right]\ ] ] where @xmath104 means the @xmath105 unit matrix .
this can be written in the equivalent form @xmath106 the simplicity of these relations , compared with for example those of ref @xcite , is related to the presence of @xmath107 in eq .
( [ chi l ] ) ( see appendices [ a ] and [ b ] ) we now turn to the scattering matrix @xmath108 for the holes . from the bdg equations we can see that the hole wavefunction @xmath109 evolves according to @xmath110 where @xmath111 is given by @xmath112 = { \cal t } h_e[+b ] { \cal t } = -h_e[-b]\ ] ] if @xmath109 describes the scattering of a hole with excitation energy @xmath113 in a field @xmath114 then @xmath115 describes the scattering of an electron at energy @xmath116 also in field @xmath114 .
we shall make use of this in two ways .
firstly , outside the disordered region @xmath117 and @xmath118 . thus outside the disordered region it is useful to expand @xmath109 in terms of the bloch states of @xmath39 at energy @xmath119 . at the left , for example @xmath120 note that since in this region @xmath118 the probability currents are reversed by comparison with the electron case .
we therefore associate the coefficient @xmath121 , the flux amplitude for a positive hole probability current in the @xmath79 direction , with the wavefunction proportional to @xmath122 .
the holes carry an opposite electric charge to that of the electrons so that they carry an electric current @xmath123 by definition , the @xmath124 matrix @xmath108 relates incoming hole probability currents to outgoing probability currents @xmath125 = \left [ \begin{array}{c } b_- \\ b_+^{\prime } \end{array } \right]\ ] ] the matrix @xmath108 has a structure similar to that of eq .
( [ se ] ) ; i.e. @xmath126\ ] ] secondly by rewriting @xmath115 in terms of electron flux amplitudes , and recalling that these amplitudes are related by @xmath90 , we arrive at a relation between @xmath127 and @xmath128 @xmath129 s_e^*(-\epsilon , + b ) \left [ \begin{array}{cc } -1_{2n } & 0 \\ 0 & 1_{2n } \end{array } \right]\ ] ] or @xmath130^ * \\ r_h^{\prime}(+\epsilon,+b ) & = - & [ r_e^{\prime}(-\epsilon,+b)]^ * \\ t_h(+\epsilon,+b ) & = + & [ t_e(-\epsilon,+b)]^ * \\ t_h^{\prime}(+\epsilon,+b ) & = + & [ t_e^{\prime}(-\epsilon,+b)]^ * \end{array}\ ] ] again we assume here a suitable choice of transverse wavefunctions ( see appendix [ a ] . ) following a similar line of argument it is possible to demonstrate a relationship between @xmath131 and @xmath132 . in this section
we outline the calculation of the coefficients of andreev reflection @xcite at the ns interface .
these are essentially unchanged by the introduction of a spin orbit interaction in the materials forming the junction .
we assume throughout that @xmath133 , a condition which is realised in practice .
far from the ns interface in the normal metal the superconducting gap @xmath134 . on the other hand , far from the ns interface in the superconductor , @xmath135 where @xmath136 is real . in general
the reflection coefficients will depend on the precise form of @xmath137 in the transition region near the junction .
for a point contact junction , however , it is permissible to assume a simple step model @xmath138 we are interested in the situation where the energy @xmath69 of the incident electron is in the energy gap of the superconductor @xmath139 . anticipating somewhat in order to avoid unnecessary algebra ,
we find the electron is mainly reflected as a hole like excitation .
a solution of the bdg equation in the normal metal corresponding to this is @xmath140 = \left [ \begin{array}{c } 1 \\ 0 \end{array } \right ] \exp ( i k_n^{(+ ) } z ) \psi_{+n}^{(+)}(x , y,\sigma ) + \left [ \begin{array}{c } 0 \\ r_{he}^a \end{array } \right ] \exp ( i k_n^{(- ) } z ) \psi_{+n}^{(-)}(x , y,\sigma)\ ] ] where @xmath141 denotes the matrix of andreev reflection amplitudes and the superscript @xmath142 refer to bloch states with energies @xmath143 .
the first term describes an excitation where an electron above the fermi level is incident from the left in channel @xmath37 .
the second term corresponds to an excitation in which an electron below the fermi level is annihilated , i.e. to a reflected `` hole '' with opposite velocity and spin to that of the incoming electron .
since @xmath144 we can to a good approximation ignore the difference between @xmath145 and @xmath146 and similarly the differences between the transverse wavefunctions .
requiring that the wavefunction and its derivative be continuous at the boundary of the superconductor leads to @xmath147 here we have made the further assumption that @xmath148 , the limit of interest in what follows , and we have also given the reflection coefficient for an incident hole . in real ns junctions , a mismatch between the conduction bands of the two materials which make up the junction results in the creation of a schottky barrier at the interface .
this barrier plays an important role in the physics of the device and so we must take it into account .
we shall model the schottky barrier as a simple planar potential barrier . at the barrier
an incident particle may be either transmitted without a change of momentum or specularly reflected .
we neglect any dependence of the reflection and transmission probabilities on the momentum of the incident particle so that the properties of the barrier are described by a single parameter @xmath149 $ ] , its transparency .
the transmission and reflection matrices which make up the electron scattering matrix @xmath150 of the barrier are @xmath151 the precise form of the @xmath98 matrix @xmath152 depends on the choice of transverse wavefunctions . for subsequent analysis
we need only note , however , that @xmath152 is antisymmetric and unitary .
the hole scattering matrix @xmath153 for the barrier is related to @xmath150 in the usual way by ( [ se vs sh ] ) .
it is the purpose of this section , having considered above the scattering matrices for the various components of the ns junction , to explain how the scattering matrices may be combined to find the total scattering matrix .
we consider first a junction without a schottky barrier .
for the normal part we can write @xmath154 \left [ \begin{array}{c } c_+ \\ c_- ^{\prime } \end{array } \right ] = \left [ \begin{array}{c } c_- \\ c_+ ^{\prime } \end{array } \right]\ ] ] where @xmath155 \;\ ; etc.\ ] ] are @xmath91-dimensional matrices and @xmath156 \ ; , & c_{\pm}^{\prime } = \left [ \begin{array}{c } a_{\pm } ^{\prime } \\ b_{\pm}^{\prime } \end{array } \right ] \end{array}\ ] ] @xmath91-dimensional vectors , in the notation of sec .
iv . for the andreev
part we define the @xmath89 scattering matrix @xmath157 by @xmath158 by reference to section iii .
this has the form @xmath159\ ] ] for the combined system , the @xmath124 @xmath160 matrix is defined by @xmath161 in equation ( [ defn s eh ] ) we replace @xmath162 in terms of @xmath163 as in ( [ defn sa ] ) ; we then perform the matrix multiplication and , from the two resulting equations , eliminate @xmath164 . from the expression relating @xmath165 and @xmath166
we extract the @xmath124 @xmath160 matrix of ( [ defn s ] ) to obtain @xmath167 in the electron - hole spaces , @xmath160 of eq .
( [ combined s ] ) has the structure @xmath168\ ] ] @xmath169 etc .
being @xmath98 matrices . to determine the conductance we shall need the submatrix @xmath170 . from eqs .
( [ combined s ] ) and ( [ s ] ) we find @xmath171 using the structure ( [ sa ] ) of @xmath157 we can write @xmath170 as @xmath172 for any nonsingular operator @xmath173 we now use the operator identity @xcite @xmath174 to find @xmath175^{-1 } t_e\ ] ] for a junction with a schottky barrier we first consider the composition of the schottky barrier and the superconductor
. the scattering matrix for this system can be obtained from eq .
( [ combined s ] ) where @xmath176 , @xmath177 and @xmath178 are taken from the model ( [ r , t b ] ) for the barrier . with the aid of the identity ( [ identity1 ] ) and
the following one @xcite @xmath179 we find @xmath180\ ] ] @xmath181 1_{2n } \\ r_{eh}^{bs } = -(r_{he}^{[bs]})^ * \\ r_{ee}^{bs } = 2 \left [ \sqrt{1-\gamma}/(2-\gamma ) \right ] w_b \\ r_{hh}^{bs } = -(r_{ee}^{[bs]})^ * \end{array}\ ] ] the scattering matrix for the complete system of normal part , schottky barrier and superconductor can now be obtained from eq .
( [ combined s ] ) , where @xmath182 , etc .
refer to the normal metal as before , but @xmath157 is replaced by @xmath183 of ( [ r bs ] ) .
( note that in deriving eq .
( [ combined s ] ) the structure ( [ sa ] ) of @xmath157 was not used . )
we assume that the bias voltage is small in comparison to @xmath184 , and that the size of the superconducting part is long enough so that there is no quasi - particle current in the superconductor . in this case
, the zero temperature dc conductance @xmath17 of the normal - superconducting junction can be described by a simplified `` landauer '' formula which has been derived in ref.@xcite @xmath185 where @xmath170 is the @xmath98 matrix of electron - hole reflection amplitudes for the composite system .
there is an important simplification if the hamiltonian of the system is time reversal invariant and the bias voltage is small in comparison to the thouless energy @xmath186 @xcite , so that the energy dependence of @xmath90 can be neglected .
the conductance then has the form @xcite
@xmath187 ^ 2\ ] ] where @xmath188 are the eigenvalues of @xmath189 .
we have verified that this result still holds when spin rotation symmetry is broken by the spin orbit interaction . in what follows
we wish to compare the quantum conductance of the ns junction , calculated from ( [ gns1 ] ) , with the classical conductance of the junction .
this latter quantity is determined using the classical rule of combining conductances @xmath190 this corresponds to the addition of flux intensities as opposed to flux amplitudes .
the conductance associated with the electron traversing the normal part is @xmath191 and similarly for the hole in the traversing the normal part in the opposite direction .
the conductance of the barrier is @xmath192 , so that the classical conductance is @xmath193 the classical conductance is insensitive to the breaking of time reversal and spin rotation symmetries .
in this section we shall calculate the conductance of an ns junction to first order in @xmath5 , where @xmath6 is the length of the normal part of the junction and @xmath7 is the elastic mean free path .
we neglect terms of order @xmath194 and above so that result is strictly applicable only in the limit that @xmath195 .
nevertheless we shall see that the results of the calculation shed considerable light on the origin of the quantum interference in the device .
the @xmath105 reflection matrix @xmath170 for the system consisting of the normal metal , barrier and superconductor can be obtained as discussed in section [ scomb ] . after expanding to second order in @xmath196 , @xmath197 , which is sufficient for an evaluation of @xmath198 to first order in @xmath199
, we have @xmath200 the conductance is found by substituting this into ( [ gns1 ] ) and performing an average over an ensemble of scattering matrices @xmath90 , describing a set of microscopically different but macroscopically equivalent configurations of impurities in the normal part of the junction . in principle , the distribution for @xmath90 should be calculated from some model distribution of hamiltonians .
we shall not , however , attempt to do that here .
instead we will assume that the resulting ensemble of scattering matrices @xmath90 is distributed according to the `` local maximum entropy model '' @xcite .
if the geometry is quasi-@xmath201 and the number of channels @xmath4 sufficiently large , then results obtained with the aid of the local maximum entropy model are known to be identical to those obtained from the class of microscopic models described by the the nonlinear sigma model @xcite .
the distribution of @xmath90 in the local model depends on @xmath4 , @xmath199 and the symmetry of the hamiltonian , i.e. whether or not time reversal symmetry is broken and whether or not spin rotation symmetry is broken .
there are _ four _ ensembles ( table .
[ table1 ] . )
the critical strengths of the magnetic field and the spin orbit interaction separating the various ensembles should be similar to those associated with the weak localisation effect in normal metals .
the details of the distribution of @xmath90 for the four ensembles can be found in appendix [ c ] .
it will be helpful , when we come to discuss the interpretation of the results , to write ( [ mss ] ) in the form @xmath202 each term in the series represents the contribution of a particular scattering processes to @xmath170 .
the classical conductance is obtained by ignoring interference between different processes and summing intensities @xmath203 the quantum correction to this classical conductance is found by summing the interference between different processes @xmath204 after carrying out the average we find that @xmath205 which agrees with the expansion of ( [ gcl ] ) to the order we are considering .
the explicit expressions for @xmath20 are collected together in table [ table2 ] .
the function @xmath206 of the barrier transparency , which appears in the table , has the explicit form : @xmath207\ ] ] there are two obvious limiting cases : @xmath208 corresponding to a junction without a schottky barrier and @xmath209 corresponding to a junction with a high schottky barrier .
the first point to note is that the quantum correction is of @xmath2 , the same order as the classical conductance , in zero field .
secondly for @xmath209 , the conductance increases as disorder is added to the junction .
this is the essence of the dramatic reflectionless tunnelling effect which we discuss in section [ reftun ] .
thirdly when time reversal symmetry is broken by the application of a magnetic field the quantum correction is @xmath9 and not @xmath13 as might have been expected by analogy with the weak localisation effect in a normal metal .
the final , and perhaps the most surprising result , is that in a magnetic field the breaking of spin rotation symmetry by the spin orbit interaction multiplies the quantum correction by a factor of minus one half even though the symmetry of the hamiltonian , in the sense of random matrix theory , is unchanged and remains unitary .
the importance of quantum interference between processes in which the path of an electron ( hole ) incident on the ns boundary is subsequently retraced by an andreev reflected hole ( electron ) was first pointed out in @xcite . in the absence of a magnetic field , and if the bias voltage is small enough , electrons which move along a path in one given sense are phase conjugated with holes traversing the time reversed path . in a magnetic field , or
if the bias voltage is large enough , this phase conjugation is destroyed .
however we have seen that there is a significant quantum correction even in a magnetic field .
there must therefore be an additional source of quantum interference which is not sensitive to the breaking of time reversal symmetry .
as we shall see the relevant processes involve paths in which an electron ( hole ) and an andreev reflected hole ( electron ) traverse a loop in the same sense . in order to remain concise
, we shall refer to such processes as containing identical paths .
the interference involving such paths can only be destroyed by applying a large enough bias voltage .
the physical importance of the bias voltage as a `` symmetry breaking parameter '' is discussed in @xcite .
only some of the terms in ( [ mss2 ] ) are found to be nonzero after averaging , so that in fact @xmath210 where @xmath211 means the @xmath212th term in ( [ mss ] . )
consider first the interference between process @xmath213 and @xmath214 .
an example of a scattering path which contributes to process @xmath213 @xmath215 is illustrated in figure 2 . since we are working to first order in @xmath199
it is sufficient to consider the motion of the electron and holes , traversing from one side of the normal part of the junction to the other , as ballistic , so these trajectories appear as straight lines in figure 2 .
examples of paths which contribute to @xmath214 @xmath216 are illustrated in figures 3 and 4 .
to first order in @xmath199 the interference between processes @xmath213 and @xmath214 is @xmath217 the remaining terms involving @xmath213 and @xmath218 contribute @xmath219 in the interest of brevity we will concentrate on the interference between @xmath213 and @xmath214 .
a very similar analysis is possible for the interference between processes @xmath213 and @xmath218 leading to identical conclusions .
we proceed by relating the product of electron and hole reflection matrices in ( [ p1p5 ] ) to a product of an electron and a hole greens function . to simplify the algebra we shall suppose that both the spin orbit interaction and the magnetic field are zero everywhere except in the disordered region .
we shall also suppose that @xmath220 and impose periodic boundary conditions in the @xmath221 direction .
the bloch states at energy @xmath71 have the form @xmath222 where @xmath223 and the energy and the momenta are related by @xmath224 the reflection matrices for electrons and holes can be related to the corresponding green s functions as indicated in appendix [ a ] .
@xmath225_{mn } & = & -i \sqrt{4\sin k_m \sin k_n } \exp(+i ( k_m+ k_n)l ) \nonumber \\ & & \sum_{x_a \sigma x_b \sigma^{\prime } } \psi_{+m}^*(x_b\sigma ) g_e^+(x_b , l,\sigma ; x_a , l,\sigma^{\prime } ) \psi_{-n } ( x_a \sigma^{\prime } ) \label{gere}\end{aligned}\ ] ] @xmath226_{nm } & = & -i \sqrt{4\sin k_m \sin k_n } \exp(-i(k_m+ k_n)l ) \nonumber \\ & & \sum_{x_a \sigma x_b \sigma^{\prime } } \psi_{-n}^*(x_b\sigma ) g_h^+(x_b , l,\sigma ; x_a , l,\sigma^{\prime } ) \psi_{+m } ( x_a \sigma^{\prime } ) \label{ghrh}\end{aligned}\ ] ] note that , for convenience , the states ( [ basis ] ) have not been normalised to carry identical currents , proper account of this has been taken in the expressions ( [ gere ] ) and ( [ ghrh ] ) for the reflection matrices .
the quantum correction involves a trace over the product of @xmath227 and @xmath228 . using the relation with the green s function this can be separated into two elements : an integration over the cross section involving the transverse wavefunctions and an average of the product of an electron and a hole greens function .
we will consider the second element first .
this involves evaluating @xmath229 within the semiclassical approximation , as is explained in ref .
@xcite , ch .
12 and ch . 13 , we can express the greens functions as summations over paths : @xmath230_{\sigma_{+m},\sigma_{-n } } \\
g_h^+(x_b , l,\sigma_{-n } ; x_a , l,\sigma_{+m } ) & = & -\sum_{j : x_a\rightarrow x_b } a_j \exp(-i s_j - i \phi_j ) [ v_j]_{\sigma_{-n},\sigma_{+m } } \end{array } \label{paths}\ ] ] substituting ( [ paths ] ) into ( [ gfavg ] ) and taking the disorder average we find that only two contributions remain : the path of the electron moving between @xmath231 and @xmath232 is retraced by the andreev reflected hole as illustrated in figure 3 .
the electron and hole charges are opposite but they traverse the path in opposite directions so the aharonov - bohm phase factors they accumulate do not cancel each other out .
thus this term is important only when the magnetic field is negligible and the hamiltonian is time reversal symmetric . let us consider first the situation where the spin orbit interaction is negligible . in the quasi-@xmath201 limit ,
the average of the product of the hole and electron green s functions is independent of @xmath231 and @xmath232 so that @xmath233 after integrating the product of transverse wavefunctions over the cross section we find @xmath234_{mn } [ r_e^{\prime}]_{nm } \right>= 4\sin k_m \sin k_n \lambda \delta_{\sigma_{-n},\sigma_{+m } } + \text{i.p . }
\label{trps}\ ] ] where i.p . denotes the contribution due to identical paths which we will consider below .
we determine @xmath235 by demanding consistency with the local maximum entropy model . in this model
@xmath90 is distributed isotropically ( see appendix [ c ] ) so that averages , such as that in ( [ trps ] ) , are independent of the channel indices @xmath37 and @xmath236 . the extent
to which the isotropy assumption is reasonable for microscopic models is discussed in refs .
@xcite . in the present case ,
to maintain consistency with the results of section [ quasiball ] , we are required to make the approximation that the group velocities appearing in ( [ trps ] ) can be replaced by an average velocity .
this is a reasonable assumption if the energy is such that we are far from any subband edges , but is of questionable validity for energies close to a subband edge .
having made this approximation and determined @xmath235 by comparison with section [ quasiball ] we find @xmath237 if the spin orbit interaction is appreciable we assume that the spin direction of an electron or hole traversing a path is completely randomised .
the matrix @xmath64 describing this rotation can then be assumed to be uniformly distributed , relative to the invariant measure , on the group @xmath62 .
the result is @xmath238 after summation over all channels the contribution of this term to the quantum correction is essentially independent of the strength of the spin orbit interaction .
this is because the spin rotation experienced by an electron traversing a path in a given sense cancels that of the hole traversing the same path in the opposite sense . to demonstrate this
explicitly it suffices to note that after summation over spin indices the contribution of time reversed paths is proportional to tr@xmath239 , which is a constant independent of the distribution of @xmath64 .
the electron path in the normal metal contains a loop which is retraced in the same sense by the andreev reflected hole as illustrated in figure 4 . the aharonov - bohm phase accumulated by the electron and hole ,
are now exactly opposite and cancel .
therefore this term is important even if time reversal symmetry is broken .
proceeding in a similar way to that above we find that the contribution of identical paths is @xmath241_{nm } [ r_e^{\prime}]_{mn } \right>= 4\sin k_m \sin k_n \lambda \delta_{m , n } \left < v_{\sigma_{-n},\sigma_{+m } } v_{\sigma_{+m},\sigma_{-n } } \right > + \text{t.r.p . }
\label{ips}\ ] ] ( here t.r.p
. refers to the contribution from time reversed paths . )
the presence of the delta function , arising from the integration of the transverse functions over the cross section , means that identical paths only contribute for electrons and holes which are back scattered by the disordered region as is indicated in figure 4 .
note the novel feature that this is true even if time reversal symmetry is broken in the disordered region .
the average over the matrices @xmath64 is calculated with respect to the distribution @xmath242 if the spin orbit interaction is negligible , and @xmath243 uniform on @xmath62 , if it is appreciable . as above
we determine @xmath235 by reference to section [ quasiball ] .
if the spin orbit interaction is negligible we find : @xmath244 in zero magnetic field and : @xmath245 in a nonzero field .
if the spin orbit interaction is appreciable we find @xmath246 in zero magnetic field and @xmath247 in a nonzero field .
the change of sign when spin rotation symmetry is broken comes about because an electron and a hole traversing the same path in the same direction undergo the same spin rotation .
after summation over the spin indices , the contribution from identical paths is thus proportional to tr@xmath248 which is sensitive to the distribution of @xmath64 @xmath249
for convenience we will defer discussion of a junction for which the scottky barrier has a low transparency ( @xmath209 ) and which exhibit reflectionless tunnelling , to section viii , and focus in this section on junctions without a schottky barrier .
we now have to sum the perturbation series ( [ mss ] ) to all orders in @xmath199 to find the quantum correction for a diffusive junction . in the absence of a magnetic field , this amounts to the evaluation of the integral : @xmath251 ^ 2 \label{gfromden}\ ] ] here @xmath252 is the density of the eigenvalues of the matrix @xmath253 , where @xmath254 refers to the composite system composed of the normal metal and schottky barrier . within the local maximum entropy model ,
the development of the density @xmath252 as a function of @xmath199 is described by a nonlinear diffusion equation @xcite .
the initial condition on @xmath252 at @xmath255 is : @xmath256 in the diffusive regime the solution to this equation is independent of the ensemble up to corrections of @xmath9 @xcite . evaluating ( [ gfromden ] ) we find an identical @xmath2 contribution to @xmath198 for the orthogonal and symplectic ensembles , while the following @xmath9 term is ensemble dependent .
for the details , we refer the reader to @xcite . with the aid of this reference
we obtain the results for the orthogonal and symplectic ensembles given in table [ table3 ] . for a diffusive junction in a magnetic field the summation of the perturbation series
requires the performance of a number of averages over the unitary group .
the authors of ref .
@xcite find for the unitary i ensemble and @xmath208 that @xmath257 this gives the same result as section [ quasiball ] , in the limit that @xmath195 and yields @xmath258 in the diffusive regime ( @xmath259 . ) as for the quasi - ballistic junction , the quantum correction is @xmath9 , not @xmath13 , in a magnetic field .
the analysis of ref @xcite can be extended to the unitary ii ensemble . in the absence of a schottky barrier we have from ( [ rhe 3 ] ) and ( [ se vs sh ] ) : @xmath260^p t_e\ ] ] then with the aid of ( [ param .
s ] ) and ( [ gns1 ] ) we obtain for the conductance : @xmath261^p u t u^{\dagger } \sum_{q=0}^{\infty } \left [ - \sqrt{r } u^t \sqrt{r } u^{\dagger } \right]^q \right]\ ] ] where @xmath262 is uniformly distributed on @xmath263 .
following ref @xcite the ensemble average is accomplished in two steps .
first the average on the unitary group and then the average over the distribution of @xmath264 .
the final result for the unitary ii ensemble and @xmath208 is : @xmath265 which reproduces the result of section [ quasiball ] for @xmath195 and yields @xmath266 in the diffusive regime .
the most interesting result is that , like the quasi - ballistic junction , the quantum correction is sensitive to the spin orbit interaction even if time reversal symmetry is broken . referring to section [ interp ] , it seems reasonable to assume that it is interference involving identical paths which is responsible for this sensitivity . in order to confirm that quantum correction is sensitive to the spin orbit interaction , even if the junction is a magnetic field we have a carried out a numerical simulation . as a model for the junction we use the hamiltonian of section [ scatter ] .
the fermi energy @xmath24 is measured from centre @xmath267 of the energy band of the model .
since the zero of energy is arbitrary we can set @xmath268 for convenience . in the normal part of the junction , @xmath269
, we take the diagonal hamiltonian elements as : @xmath270 the potential @xmath271 is the random potential due to impurities and @xmath272 the height of the schottky barrier . for simplicity we shall assume a spatially uncorrelated potential with a distribution @xmath273 @xmath274 to the right , in @xmath275 , is the superconductor .
the conductance is calculated using the standard green s function iteration technique to produce @xmath90 and then using the formulae of sections [ scatter ] and [ formulae ] .
in figure 5 we present the data for the junction conductance versus the parameter @xmath56 , defined in section [ scatter ] , controlling the spin orbit interaction .
the transverse dimensions of the junction are @xmath276 and the longitudinal dimension @xmath277 . the fermi energy @xmath278 . for these parameters @xmath279 .
we are interested in the limit @xmath208 so we set @xmath280 .
a magnetic field corresponding to a flux of @xmath281 per lattice cell is applied perpendicular to the junction corresponding to five flux quanta through the device .
the distribution of the random potential has width @xmath282 corresponding to an estimated mean free path @xmath283 .
the junction is in the diffusive ( @xmath284 ) quasi-@xmath201 ( @xmath285 , @xmath286 ) regime
. the spin rotations between lattice sites are not random but differ depending on the direction of the displacement .
since these rotations do not commute the spin direction of the electron is effectively randomised and so , for @xmath56 large enough , the symplectic ensemble should be appropriate .
as the magnitude of the spin orbit interaction is increased the junction conductance @xmath17 increases , while the normal conductance @xmath287 remains constant .
this is exactly the behaviour predicted .
the value @xmath288 obtained in the simulation is also in reasonable agreement with theoretical value @xmath289 .
this is an effect observed in diffusive junctions when @xmath15 . in the absence of a magnetic field
we may calculate the conductance in the same way as for the diffusive junction with @xmath208 but with the initial condition : @xmath290 instead of ( [ initcond ] ) . at @xmath2
the result is the same for both the orthogonal and symplectic ensembles : @xmath291 we expect ensemble dependent corrections to this result at @xmath9 , though we have not calculated them explicitly .
this quantum conductance has to be compared to the classical conductance obtained from eq .
( [ gcl ] ) by setting @xmath292 : @xmath293 for both the orthogonal and symplectic ensembles the conductance is enhanced above the classical value , with a correction @xmath2 .
note in particular that in the absence of a field the conductance increases as disorder is added to the junction .
when the length of the junction is such that @xmath16 , the junction conductance reaches its maximum value of roughly @xmath294 .
the classical value at this length is @xmath295 .
this very large enhancement of the conductance is referred to as `` reflectionless tunnelling . '' in practice it is the sensitivity of the reflectionless tunnelling effect to a magnetic field which is observed in experiments @xcite .
we have seen explicitly in the analysis of the quasi - ballistic junction that the phase conjugation of electrons and holes traversing time reversed paths is destroyed in a magnetic field , and only the @xmath9 correction , due to interference involving identical paths , should remain .
thus , by applying a magnetic field , the conductance of the junction should be dramatically reduced to its classical value , within a correction of @xmath296 .
this argument also indicates that if we break spin rotation symmetry , the qualitative behaviour in a magnetic field should be unchanged , and there should only be quantitative corrections at @xmath9 due to the spin orbit interaction . to confirm this
we have carried out a numerical simulation of a junction under the relevant conditions .
the model used has been described in section [ diffjunc ] .
the parameters used in the simulation are @xmath297 , @xmath298 and @xmath278 for which @xmath299 at @xmath300 and @xmath301 at @xmath302 .
we have set @xmath303 and estimate that the mean free path @xmath304 .
the junction is in the diffusive ( @xmath284 ) quasi-@xmath201 ( @xmath285 , @xmath286 ) regime . in figure 6
we present the data for the conductance versus magnetic field obtained in the simulations for the two values of spin orbit interaction parameter @xmath56 indicated above .
the magnetic field is given in terms of the flux @xmath305 penetrating the normal part of the junction . at z
= l there is a potential barrier of height @xmath306 .
the barrier transparency is a function of @xmath272 and also has a weak dependence on the the spin orbit parameter @xmath56 and the magnetic flux @xmath305 .
the average value of @xmath307 with a variation of about @xmath308 over the parameter range used . at zero field and zero bias
the conductance should be approximately : @xmath309 which is in reasonable agreement with the numerical data .
on applying a magnetic field we expect to approach the classical value : @xmath310 this is again in reasonable agreement with the numerical data .
a strong reduction of the conductance is observed both when spin rotation symmetry is unbroken @xmath300 and when it is broken @xmath311 by the spin orbit interaction .
the critical field corresponds to a flux @xmath312 penetrating the sample , in agreement with critical field expected for the suppression of the weak localisation effect in normal metals .
the data confirm that , apart from possible corrections of @xmath9 , the reflectionless tunnelling effect is not sensitive to the spin orbit interaction .
we have studied the effect of the breaking of spin rotation symmetry , as a result of an appreciable spin orbit interaction , in determining the quantum correction to the classical conductance of a disordered mesoscopic normal superconducting junction .
the most striking result we have obtained concerns the ns junction in an applied magnetic field . we have found that even if time reversal symmetry is broken , the quantum correction is sensitive to the spin orbit interaction .
a semiclassical analysis of a quasi - ballistic junction permitted the scattering processes , which interfere to produce this effect , to be identified .
they involve scattering paths in which an electron and a hole traverse a loop in the normal part of the junction in the same sense , that which we have called here `` identical paths '' .
keith slevin would like to thank the european commission for financial support under the human capital and mobility programme .
pier mello would like to thank the wissenschaftskolleg zu berlin for its hospitality , as well as the service de physique de ltat condens , cea - saclay , for the financial support given during his visit to saclay in the early stages of this research .
we would also like to thank piet brouwer for his helpful comments . in this appendix
we group together certain technical details concerning the scattering theory for model ( [ model ] ) which appears in the main text .
we adopt the dirac notation so that , for example , @xmath313 . the particle current @xmath314 through a cross section at @xmath48 can be shown to be : @xmath315 | \psi>\ ] ] where @xmath173 is the operator : @xmath316 & = & \sum_{xy\sigma\mu } |xyz\sigma><xyz\sigma|h_e|xyz+1\mu><xyz+1\mu| \nonumber \\ & & -|xyz+1\sigma><xyz+1\sigma|h_e|xyz\mu><xyz\mu|\end{aligned}\ ] ] by rewriting the schroedinger equation in the contact in terms of a real space transfer matrix , and considering the orthogonality relations between left and right eigenvectors of this matrix , it is possible to show that the bloch states satisfy the orthogonality relation : @xmath317|n > = 0 \text { if } k_m\neq k_n^ * \label{orthog}\ ] ] in addition for real @xmath318 the transverse wavefunctions may be normalised so that : @xmath319|\pm n > = \pm i \delta_{n , m } \label{norm}\ ] ] each channel then carrys a charge current of @xmath320 . here
we use the notation that : @xmath321 in the absence of a magnetic field the hamiltonian @xmath39 will commute with the time reversal operator @xmath322 . in this case if @xmath33 is a solution to the scattering problem we may construct another solution @xmath323 by operating on @xmath70 with @xmath29 . taking note of this
we can derive the following condition on @xmath90 if the hamiltonian is time reversal invariant : @xmath324 s_e \left [ \begin{array}{cc } -d & 0 \\ 0 & d \end{array } \right]\ ] ] where @xmath325 is a @xmath98 symmetric unitary matrix with elements : @xmath326_{\sigma,\mu } \exp(i k_m ) - [ v^z]_{\sigma,\mu } \exp(-i k_n ) \right ] \psi_{+n}^*(x , y,\mu)\ ] ] the condition on the scattering matrix when @xmath39 is time reversal invariant , can be simplified by an appropriate choice of basis . since @xmath325 is symmetric unitary it may be decomposed ( not uniquely ) as @xmath327 with @xmath328 unitary . if we do this and make a transformation to the new basis : @xmath329 the matrix @xmath325 becomes unity @xmath330 .
this transformation is legitimate since it leaves the current normalisation unchanged .
the general relation between the electron and hole scattering matrices arrived at is , as indicated in the main text : @xmath331 s_e^*(-\epsilon,+b ) \left [ \begin{array}{cc } -d^ * & 0 \\ 0 & d^ * \end{array } \right]\ ] ] which again may be simplified by the transformation ( [ newbasis ] ) .
we now turn to the relation between the electron scattering matrix @xmath90 and the electron green function @xmath332 .
let us suppose that @xmath333 with @xmath334 a random potential .
to each incoming state at the left @xmath335 we associate a scattering state @xmath336 : @xmath337 supposing that the random potential @xmath334 is nonzero only in the volume @xmath338 between @xmath339 this can be transformed to : @xmath340 |+n > - g^+_e d[h_e , z_0 - 1 ] |+n >
\nonumber \\ & & + \sum_{xyz\notin \cal{v } \sigma } |xyz\sigma><xyz\sigma|+n>\end{aligned}\ ] ] in @xmath341 the scattering state has the form : @xmath342_{m , n } |-m>\ ] ] and in the region @xmath343 : @xmath344_{m , n } |+m>\ ] ] with the aid of ( [ orthog ] ) and ( [ norm ] ) we now find : @xmath345_{mn } & = & -i
< -m| d[z_0 ] g^+_e d[z_0 - 1 ] |+n > \\
\left[t_e\right]_{mn } & = & + i < + m| d[z_1 - 1 ] g^+_e d[z_0 - 1 ] |+n > \\
\left[r^{\prime}_e\right]_{mn } & = & -i < + m| d[z_1 - 1 ] g^+_e d[z_1 ] |-n > \\ \left[t^{\prime}_e\right]_{mn } & = & + i < -m| d[z_0 + 1 ] g^+_e d[z_1 ] |-n > \end{array}\ ] ] the last two relations are obtained by considering an incoming wave @xmath346 from the right . in a similar way a relation between @xmath108 and @xmath347 can be derived .
we note only the expression for the reflection matrix at the right : @xmath348_{mn } = -i < -m| d[z_1 - 1 ] g^+_h d[z_1 ] |+n>\ ] ] and we now assume that the magnetic field is zero except in @xmath338 . as mentioned in the text , the simplicity of ( [ se vs sh ] ) is related to the presence of @xmath107 in eq.([chi l ] ) . in most references dealing with random matrix approaches to conduction in disordered solids definitions , such as those of ref .
@xcite , @xmath107 does not appear in the equivalent of these equations . to avoid confusion
we describe explicitly the transformation between the sets of definitions . in the absence of a magnetic field , the scattering matrix of ref .
@xcite which we shall call @xmath349 ( containing reflection and transmission matrices @xmath350 , @xmath351 , etc . ) has the form : @xmath352\ ] ] and the property that : @xmath353 s_e^{mp } \left [ \begin{array}{cc } z & 0 \\ 0 & z \end{array}\right]\ ] ] where the @xmath105 antisymmetric unitary matrix @xmath354 satisfies ( [ z1 ] ) and ( [ z2 ] ) .
the coefficients @xmath355 , @xmath356 introduced in section iv are identical to those of ref .
@xcite , while the coefficients @xmath357 , @xmath358 are related by : @xmath359 the relation between @xmath160 matrices is then : @xmath360 s_e^{mp } \left [ \begin{array}{cc } 1 & 0 \\ 0 & z \end{array}\right]\ ] ] so that ( [ smp vs smpt ] ) and ( [ s vs smp ] ) give the relation ( [ s vs st ] ) . in the absence of spin - orbit coupling the two spin components
would be independent and the reflection and transmission matrices would be block diagonal so that , in a suitable basis , @xmath361 would have the structure : @xmath362\ ] ] here , @xmath363 , @xmath364 , @xmath365 are @xmath366 matrices that correspond to the reflection and transmission matrices for either spin direction and have the usual properties , so that @xmath349 is symmetric .
the @xmath160 of this paper , obtained from the transformation ( [ s vs smp ] ) , would have the structure : @xmath367\ ] ] the parameterisations of @xmath90 appropriate for the various ensembles are discussed in refs .
@xcite . for the unitary ii ensemble @xmath90
may be parameterised : @xmath368\ ] ] where the @xmath369 s are @xmath98 unitary matrices , @xmath370 and @xmath371 .
( the parameters @xmath188 appearing here are the same as the eigenvalues of @xmath189 appearing in the conductance formulae of section vi . )
the unitary matrices are independently and uniformly distributed with respect to the invariant measure on the group @xmath263 of @xmath105 unitary matrices . in the symplectic ensemble ,
time reversal symmetry imposes further restrictions on the form of @xmath90 and the appropriate parameterisation is : @xmath372\ ] ] with @xmath373 and @xmath374 uniformly distributed on @xmath263 and @xmath354 is a fixed antisymmetric @xmath98 matrix satisfying : @xmath375=0 \label{z1}\ ] ] @xmath376 if the spin orbit interaction is negligible the spin is conserved during the electron motion and the hamiltonian and scattering matrices can be block diagonalised .
thus in the unitary i ensemble a basis can be found such that : @xmath377\ ] ] where @xmath378 are @xmath366 unitary matrices uniformly distributed on the unitary group @xmath379 and @xmath380 and @xmath381 .
the distribution of the @xmath188 s is given by the solution of a fokker planck equation , the precise form of which depends on the ensemble under consideration . the joint distribution , and in particular the correlations , of the @xmath188 s depend on the symmetry of the ensemble under consideration . however , all we need to know for the present purpose is that : @xcite @xmath385 independent of the ensemble .
the only further information needed to carry out the required averages is that if @xmath369 is an @xmath366 unitary matrix uniformly distributed on @xmath379 then @xmath386 an important property of the local maximum entropy model is that the distribution of @xmath90 is isotropic , that is to say the matrices @xmath387 are distributed according to the invariant measure on the unitary group and independently of the parameters @xmath388 .
g. bergmann , physics reports * 107 * , 1 ( 1984 . ) g. bergmann , solid state communications * 42 * , 815 ( 1982 . ) k. slevin and t. nagao , international journal of modern physics b * 9 * , 103 ( 1995 . ) a. kastalsky , a. kleinsasser , l. greene , r. bhat , f. milliken and j. harbison , physical review letters * 67 * , 3026 ( 1991 . ) b. van wees , p. de vires , p. magnee and t. klpawijk , physical review letters * 69 * , 510 ( 1992 . ) the random phase or isotropy assumption for the scattering matrix is described in appendix [ c ] and in refs .
@xcite y. takane and h. otani , journal of the physical society of japan * 63 * , 3361 ( 1994 . ) c. w. j. beenakker , physical review b * 49 * , 2205 ( 1994 . ) c. beenakker in _ mesoscopic quantum physics _ , edited by e. akkermans , g. montambaux , j - l pichard and j. zinn - justin ( north - holland , amsterdam , to be published . ) p. brouwer and c. beenakker , physical review b,*52 * , r3868 ( 1995 . ) p. de gennes , _ superconductivity of metals and alloys _ ,
( benjamin , new york , 1966 . ) t. ando and h. tamura , physical review b * 46 * , 2332 ( 1992 . )
g. arfken , _ mathematical methods for physicists _ , ( academic press , new york 1970 . ) n. zanon and j - l .
pichard , journal de physique france * 49 * , 1387 ( 1988 . ) s. n. evangelou and t. ziman , journal of physics c * 20 * , l235 ( 1987 . ) p. mello and j - l pichard , journal de physique 1 france * 1 * , 493 ( 1991 . ) a. f. andreev , soviet physics jetp * 19 * , 1228 ( 1964 . ) h. feshbach , _ topics in the theory of nuclear reactions _ , in _ reaction dynamics _
( gordon and breach , new york , 1973 ) .
g. e. blonder , m. tinkham and t. m. klapwijk , physical review b * 25 * , 4515 ( 1982 . ) c. lambert , journal of physics : condensed matter * 3 * , 6579 ( 1991 . ) y. takane and h. ebisawa , journal of the physical society of japan * 61 * , 1685 ( 1992 . ) for a diffusive system the thouless energy is @xmath389 where @xmath173 is the diffusion constant and @xmath6 a typical dimension of the sample . for a ballistic sample @xmath390 where @xmath391 is the fermi velocity . c. beenakker ,
physical review b * 46 * , 12841 , ( 1992 . )
p. mello , p. pereyra and n. kumar , annals of physics * 181 * , 290 ( 1988 . )
a. d. stone , p. a. mello , k , a .
muttalib and j - l .
pichard , in _ mesoscopic phenomena in solids _ , edited by b. l. altshuler , p.a .
lee and r. a. webb ( north holland , amsterdam , 1991 . ) k. frahm , physical review letters * 74 * , 4706 ( 1995 . ) p. brouwer and k. frahm , to appear in phys .
b. m. gutzwiller , _ chaos in classical and quantum mechanics _ , ( spring verlag , new york 1990 . ) k. slevin , j - l .
pichard and k. a. muttalib , journal de physique 1 france * 3 * , 1387 ( 1993 . ) r. a. jalabert and j - l .
pichard , journal de physique 1 france * 5 * , 287 ( 1995 . ) p. a. mello and j .- l .
pichard , physical review b * 40 * , 5276 ( 1989 . ) c. beenakker , b. rejaei and j. melsen , physical review letters * 72 * , 2470 ( 1994 . ) h. u. baranger and p. a. mello , physical review letters * 73 * , 142 ( 1994 ) r. a. jalabert , j .- l . pichard and c. w. j. beenakker , europhysics letters * 27 * , 255 ( 1994 . ) p. mello and a. stone , physical review b * 44 * , 3559 ( 1991 . ) a. macedo and j. chalker , physical review b * 46 * , 14985 ( 1992 . )
figure 2 : an example of a path which contributes to process @xmath213 corresponding to an electron ( solid line ) traversing the normal part of the junction whose path is then retraced by the andreev reflected hole ( dashed line ) .
figure 3 : an example of a `` time reversed path '' which contributes to process @xmath214 .
the path of the electron moving from @xmath231 to @xmath232 is retraced by the andreev reflected hole as it moves from @xmath232 to @xmath231 .
interference between this path and that illustrated in figure 2 is insensitive to the spin orbit interaction and is suppressed in a magnetic field .
figure 4 : an example of an `` identical path '' which contributes to the process @xmath214 .
the electron moves around a loop and returns to @xmath231 .
the andreev reflected hole traverses the loop in the same direction .
interference between this path and that illustrated in figure 2 is sensitive to the spin orbit interaction but not to a magnetic field .
figure 5 : the conductance of a quasi-@xmath201 diffusive ns junction as a function of the spin orbit interaction parameter @xmath56 in the presence of an applied magnetic field ( flux of @xmath392 per lattice cell ) .
the barrier transparency @xmath208 .
the ns conductance is sensitive to the breaking of spin rotation symmetry even though time reversal symmetry is broken , while the normal conductance is approximately constant as expected .
figure 6 : the conductance of a quasi-@xmath201 ns junction as a function of magnetic flux penetrating the normal part of the junction for zero @xmath300 and strong @xmath311 spin orbit interaction .
the barrier transparency is low ( @xmath393 ) and the length of the junction is such that @xmath16 .
we see that the qualitative features of the reflectionless tunnelling effect are insensitive to the spin orbit interaction . | we calculate the quantum correction to the classical conductance of a disordered mesoscopic normal - superconducting ( ns ) junction in which the electron spatial and spin degrees of freedom are coupled by an appreciable spin orbit interaction .
we use random matrix theory to describe the scattering in the normal part of the junction and consider both quasi - ballistic and diffusive junctions .
the dependence of the junction conductance on the schottky barrier transparency at the ns interface is also considered .
we find that the quantum correction is sensitive to the breaking of spin rotation symmetry even when the junction is in a magnetic field and time reversal symmetry is broken .
we demonstrate that this sensitivity is due to quantum interference between scattering processes which involve electrons and holes traversing closed loops in the same direction .
we explain why such processes are sensitive to the spin orbit interaction but not to a magnetic field .
finally we consider the effect of the spin orbit interaction on the phenomenon of `` reflectionless tunnelling . ''
short title : the so interaction in mesoscopic ns junctions . |
studies of nucleon decay ( see , _ e.g. _ , @xcite ) are among the most important in physics , in that they provide direct tests of fundamental theories of particle physics .
the different particle theories make rather different predictions as to what the nucleon decay half - lives might be , and even which decay modes would be expected to dominate .
although a number of searches for nucleon decay have been performed in the large detectors that exist , no convincing evidence has yet been presented for its existence .
the best limits from these studies have involved decay modes in which protons decay to relativistic charged leptons ( see , _ e.g. _ , @xcite ) , either as individual protons or as protons in nuclei , as the resulting cherenkov radiation would produce definitive signatures , typically of order 10@xmath2 y. however , nucleon decays in composite nuclei @xcite might produce clear evidence for the existence of such effects that could not be obtained from decay of isolated protons .
furthermore , these might allow extension of the existing limits of the decay branches in some instances , even with relatively small detectors .
in this study we focus on the n @xmath3 @xmath4 decay , which has a less well - known half - life because its decay can often be masked by backgrounds , and would in any event be difficult to detect via the means used to observe nucleon decay into relativistic leptons .
it is not necessarily expected to be the dominant mode of decay , but it is the least - well - determined mode , so is the primary limitation to the mode - independent half - life . moreover , there are non - standard models in which unusual decay modes may turn out to be the dominant ones ( see , _ e.g. _ , @xcite ) . in some models
the three neutrino decay mode can emerge naturally as the dominant one @xcite .
decays of the three neutrino mode were searched for a decade ago , @xcite , then their limit was improved to its current experimental value of 5x10@xmath5 y @xcite .
recent suggestions for studying this decay mode have included signatures that would result from decay of a neutron in the o in h@xmath6o @xcite of super - kamiokande and in the c in the ch@xmath6 @xcite of kamland .
these suggestions both involve signals that would be generated by rather weak branches resulting from nucleon decay .
in this paper we study two nuclides , @xmath0pb and @xmath1cl which , we show , would produce special responses to neutron decay .
we present the signatures that might result in the n @xmath3 @xmath7 decay .
we find that @xmath0pb would have good sensitivity to this mode , but @xmath1cl , while not producing as strong a limit , would exhibit an unusually definitive signature .
both nuclei are part of a planned detector , lead perchlorate , lpc , pb[clo@xmath8@xmath6 , that will be part of omnis , the observatory for multiflavor neutrinos from supernovae @xcite .
lpc is a colorless liquid that is highly soluble in water .
the properties of such a mixed liquid have been studied extensively @xcite ; it was found that the lpc will produce cherenkov light from relativistic particles , e.g. , electrons and muons , that might be produced in many nucleon decays .
in addition , each neutron produced by neutron emission from a nucleus within the detector will be captured on the @xmath1cl within tens of @xmath9s , producing 8.6 mev of @xmath10-rays .
an energetic @xmath10-ray will also produce cherenkov light , with a pattern that is indistinguishable from that from the relativistic leptons in lpc .
in general we can write the following expression for the neutron decay lifetime , @xmath11 , as @xmath12 where n@xmath13 is the number of neutrons in , _ e.g. _ , @xmath0pb or @xmath1cl , @xmath14 is the fraction of those that can decay into detectable signatures , r@xmath15 is the observed event rate , and @xmath16 and @xmath17 are the detection efficiencies for the two ( or more ) signatures of the decay .
the factor @xmath18 accounts for the possibility of multiple neutron emission from decay of a neutron in @xmath0pb , each with detection efficiency @xmath17 .
the factor br is the branching ratio for the decays that go to the specific decay mode being studied . in @xmath0pb decay , this could , _
e.g. _ , refer to the branching ratio for producing one neutron and a subsequent @xmath10-ray with sufficient energy to be observed ; we have assumed e@xmath19 @xmath20 3 mev in the daughter nucleus for this criterion to be satisfied , as virtually every level with excitation energy above that will produce at least one 3 mev @xmath10-ray @xcite . in @xmath1cl decay , br refers to a decay to states that will produce a @xmath10-ray with at least 3 mev of energy together with decay to the @xmath21cl ground state .
( @xmath21cl has an isomeric state at 0.146 mev , to which roughly half of the highly excited states will ultimately decay .
however , it has a much longer half - life than the 1.5 s half - life of the ground state , so will be assumed not to be useful for the present discussion . ) in order to estimate the probability of observing decays from @xmath0pb and @xmath1cl we will have to consider the probability of population of sufficiently highly excited states in the daughter nuclei : @xmath22pb and @xmath21cl .
generally the more deeply bound the nucleon that decays the larger will be the excitation energy in the daughter nuclide .
the nuclide that decays will be a nuclide in the parent nucleus , so the resulting state " of the initial nucleus , @xmath0pb or @xmath1cl , minus one neutron will map onto actual states in the daughter nuclide ( assuming the neutron decay leaves the resulting nucleus intact ; a reasonable assumption for the decay mode that produces three particles that interact only through the weak interaction ) .
however , consider the energy @xmath23 of the neutron decay products ( e.g. the 3 neutrinos ) when the final nucleus a-1 is excited to the excitation energy @xmath24 .
then @xmath25 , where @xmath26 is the neutron separation energy in the parent nucleus a. the energy in parentheses is just the binding energy of the neutron that decayed .
then , all other things being equal , extremely large values of @xmath24 will be suppressed by the decrease of the phase space of the 3 neutrinos .
this will tend to favor the excitation energies of the states in the daughter being similar to the binding energy of the neutron that decayed .
of course , this argument also assumes that the recoil kinetic energy of the residual nucleus will be negligible .
as discussed above , the signature of this nucleon decay would be given by the specific properties of the residual nuclei resulting from the decay .
in this case , @xmath0pb would first become @xmath22pb .
what would happen next would depend on the excitation energy of the states in @xmath22pb that were populated compared to the one - neutron emission threshold , the two - neutron emission threshold , etc .
, in @xmath22pb .
subsequent neutron emissions will be much more rapid than electromagnetic de - excitations , so subsequent neutron emissions would occur instead of de - excitations of the daughter nuclei as long as the states populated were above the neutron - decay threshold . the excitation energy in the post - decay @xmath22pb will depend on the binding energy of the nucleon that decayed .
this energy would be expected to be as much as several tens of mev in lead , but would be weighted toward lower values by the preponderance of higher spin nuclear orbits , hence higher occupation numbers , near the fermi surface . thus a reasonable range to assume for the excitation energy in @xmath22pb might be an asymmetric distribution ranging from essentially zero ( if a valence neutron decayed ) to as much as 20 mev , with an even higher - energy tail of the distribution extending to several more tens of mev , but with the highest energies suppressed .
the one - neutron ( two - neutron ) emission threshold in @xmath22pb is 6.74 mev ( 14.83 mev ) .
decays of neutrons in the highest - energy occupied shells would presumably primarily populate states below the one - neutron - emission threshold in @xmath22pb .
although these states would decay by emitting @xmath10-rays , @xmath22pb is stable .
thus the @xmath10-rays would be the only signature of the neutron decays to the low - lying states , providing a less - than - compelling signature of nucleon decay .
however , a fairly large fraction of the neutron decays in @xmath0pb would produce at least one neutron emission .
these states would produce @xmath28pb plus j neutrons , with j being at least one .
the branchings into the particle and @xmath10-ray emission channels at a given excitation energy of the daughter nucleus @xmath22pb have been calculated .
the relevant transmission coefficients were determined with the same inputs used in the non - smoker statistical model code @xcite , which is often used for astrophysical calculations , and has been found to be accurate over a wide mass range @xcite .
in addition to single - particle emission , two - particle emission can be calculated , specifically for the case at hand , the two - neutron emission . for two - particle emission each transition was followed from a level with given spin and parity in @xmath22pb to a definite level with given spin and parity in @xmath29pb , then probabilities for subsequent neutron emission were determined by summing over neutron emission to all possible final states in @xmath30pb . up to 20 low - lying experimentally known levels were used in each nucleus involved , and a theoretical level density @xcite was employed above the last known state .
the relative probability of one- and two - neutron emission as a function of excitation energy in @xmath22pb is shown in fig.[fig : pb207 ] .
spins from @xmath31 to @xmath32 and both parities were considered in @xmath22pb .
the transitions to the states of different spin were weighted only with @xmath33 . in order to obtain the correct decay probability , the calculated probabilities have to be folded with the function that describes how the @xmath22pb levels of a given excitation energy
are populated in the primary decay event , as described above .
the basic features of the present results will remain although the actual distribution of strength would be expected to be distorted somewhat due to the folding with the population derived from @xmath34 decay . as can be seen from fig .
[ fig : pb207 ] , the neutron channel dominates the possible decays above the neutron emission threshold . only up to about 9 mev is the decay governed by photon emission .
although the proton emission threshold is at 7.5 mev , there is no significant proton emission up to 40 mev in excitation due to the high coulomb barrier ; it is still only 20% of all decays at 48 mev of excitation energy .
although the two - neutron emission channel opens at 14.83 mev , it remains insignificant up to 30 mev in excitation energy reaching 40% of all transitions at 48 mev .
it should be noted that the solid curve in fig .
[ fig : pb207 ] contains the sum of one- and two - neutron emission .
therefore , emissions of single and double neutrons contribute equally at the highest calculated energy , each comprising 40% of all decay possibilities .
thus the population of excitation energies close to the neutron separation energy can be probed by measuring @xmath10-rays , whereas higher excitation energies are accessible by two - neutron decay .
the lowest lying 82 neutrons in @xmath0pb would be sufficiently strongly bound that their decay would be very likely to populate states at least above the one - neutron emission threshold .
some of the remaining 44 neutrons would also be likely to produce daughter states above that threshold .
their ordering is h@xmath35 , f@xmath36 , f@xmath37 , p@xmath38 , i@xmath39 , and p@xmath40 , and their approximate binding energies are 12 , 11 , 8 , 8 , 8 , and 7 mev respectively @xcite .
it seems reasonable to assume that decay of the neutrons in the last three orbitals would populate states below the one - neutron emission threshold .
thus 106 of the 126 neutrons in @xmath0pb would be expected to populate states above the one - neutron emission threshold in @xmath22pb .
an additional concern , though , is that the decays of neutrons in the most tightly bound states ,
_ i.e. _ , those dominated by the 1s@xmath40 , 1p@xmath38 , 1p@xmath40 , 1d@xmath37 , 2s@xmath40 , and 1d@xmath38 orbitals , might result in states in @xmath22pb that would decay by proton or @xmath41-particle emission .
their binding energies are 40 , 35 , 35 , 32 , 31 , and 29 mev respectively @xcite . however , our emission probability calculations described above suggest that only the states having 1s@xmath40 neutrons would be expected to exhibit such decays ; we have subtracted the two neutrons in that orbital from consideration .
thus 104 neutrons in @xmath0pb can decay to produce detectable signatures .
the upshot is that this nucleon decay mode would have a high probability , roughly 104/126 ( so that @xmath14 ( see eq .
1 ) is 104/126=0.825 ) , for producing at least one neutron in coincidence with a fairly high - energy @xmath10-ray , but emission of several neutrons in coincidence with a @xmath10-ray would also have a relatively high probability .
we have assumed 50% for the sum of the probabilities of events in which at least one neutron and a detectable ( @xmath42 3 mev )
@xmath10-ray are emitted .
we note that states above 4 mev of excitation in @xmath22pb have a high probability , at least 50% @xcite , of producing a @xmath10-ray of at least 3 mev in energy .
we assume 1 kt of lead perchlorate admixed with 20% water ( 0.41 kt of lead ) ; that has 5.9x10@xmath43 @xmath0pb nuclei , or 6.1x10@xmath44 neutrons in @xmath0pb that would be expected to populate states above the one - neutron emission threshold of @xmath22pb and decay subsequently by neutron emission .
if the fraction of the 104 neutrons assumed that decay by single or multiple neutron emission is 100% , the probability for detection of at least one neutron is 50% ( it will be considerably higher for multiple neutron emission ) , the probability for producing a detectable @xmath10-ray is 50% and its detection efficiency is 50% , then the probability that a neutron decay event from one of the 104 neutrons assumed to be detectable will be observed is 12.5% . if the lifetime for this decay branch is 10@xmath45 years then , under these circumstances , 7 - 8 events per year would be observed .
note that we have assumed that only the @xmath0pb would contribute ; it is likely that the other lead isotopes would produce similarly detectable decays , producing up to a factor of 2 enhancement .
furthermore , the assumed efficiencies are conservative .
thus this mode could have its lifetime extended by searching for these decays in lead , provided the background events could be managed .
the most obvious background signal that has several neutrons that would nt be vetoed by a cosmic ray shield would involve production of those neutrons by an energetic cosmic ray in the surrounding rock .
one might then get a @xmath10-ray if one of the neutrons could inelastically excite a nucleus .
however , this background could be eliminated easily , as the lpc detector could be surrounded by moderator , so that none of the neutrons getting into it would have enough energy to inelastically excite a nucleus .
a more serious background would result from neutrinos produced in the earth s upper atmosphere inelastically exciting @xmath0pb via the neutral - current interaction to levels from which it could emit one or more neutrons , going to nuclei that might themselves emit more neutrons .
this background would be impossible to reject on an event - by - event basis .
however , its yield could be estimated as a function of the energy of the incident neutrino by measuring the number of charged - current interactions as a function of energy ( as determined by the energy of the recoiling lepton ) that would produce similar numbers of neutrons in coincidence with the @xmath10-ray .
one could then infer the yield from the neutral - current interactions on the basis of the relative magnitudes of the two types of cross sections and the energy distribution of the charged - current interactions .
the abundance of @xmath1cl is 75% of natural cl , and there are 2 cl atoms per pb atom in lpc .
thus it is also useful to see if nucleon decay in @xmath1cl might produce a definitive signature .
neutron decay in @xmath1cl would sometimes produce @xmath21cl in a highly excited state .
however , the one - proton emission threshold for @xmath21cl is at 5.14 mev , far below the one - neutron emission threshold at 11.51 mev . indeed , there are apparently no bound levels in @xmath21cl even close to 11.51 mev .
thus the result of neutron decay to @xmath21cl will be @xmath10-ray decays to the ground state of @xmath21cl .
note , though , that @xmath21cl has an isomeric state at 0.146 mev ; so roughly half of the energetic @xmath10-rays would go to that state .
the relative branchings into different decay channels of @xmath21cl were calculated using the same methods as described above for @xmath22pb .
the results are shown in fig .
[ fig : cl35 ] .
the situation for @xmath21cl is different than for @xmath22pb , however , because of the lower coulomb barrier and a proton separation energy that is lower than the neutron separation energy .
neutron emission is relatively unimportant at all calculated excitation energies . the excited nucleus @xmath21cl will de - excite via @xmath10-transitions for excitation energies up to about 9.5 mev , where proton emission accounts for 50% of all emissions .
proton emission quickly rises to nearly 100% by 11.9 mev and the falls off at high energies due to increased @xmath41 emission .
two - neutron emission is completely negligible at all calculated energies . as with @xmath22pb , a quantitative description of the emission requires the knowledge of the population of the excited states by the decay , but would not be expected to differ qualitatively from the results shown here .
the ground state of @xmath21cl , however , @xmath46-decays with a half - life of 1.53 s. thus a coincidence between the energetic @xmath10-rays ( assumed detection efficiency = 50% ) and the @xmath47-decay would identify a candidate event for neutron decay in @xmath1cl .
the end point energy of the @xmath47 is 4.47 mev , so most of the @xmath47s will be produced with sufficient energy to be detected from their cherenkov radiation ( assumed detection efficiency = 70% ) .
the isomeric state has a much longer half - life ( 32 m ) , so we have assumed that would not produce a useful coincidence for determining nucleon decay ( reducing the useful event yield by half ) . none the less , nucleon decay in @xmath1cl could be identified by observing @xmath46 and @xmath10s in coincidence , with enough events detected to confirm the half - life of @xmath21cl . the relatively long half - life of @xmath21cl would demand that the materials used in omnis lpc modules be as pure as possible to minimize accidental coincidences . again assuming 1 kt of lead perchlorate admixed with 20% water ,
there will be 1.8x10@xmath45 @xmath1cl nuclei .
of the 18 neutrons in @xmath1cl those in the 1s@xmath40 , 1p@xmath38 , and 1p@xmath40 orbits ( binding energies = 36 , 27 , and 23 mev respectively @xcite ) , would probably decay to sufficiently highly - excited states in @xmath21cl that they would emit a proton , neutron , or @xmath41-particle , and not end up in @xmath21cl .
the remaining three orbitals , d@xmath37 , s@xmath40 , and d@xmath38 have binding energies of roughly 17 , 14 , and 10 mev @xcite .
the one - neutron separation energy in @xmath1cl is 12.64 mev ; removal of a valence neutron , presumably from a d@xmath38 dominated state , would require that much energy , and would tend to populate the ground state of @xmath21cl .
thus , presumably , decay of neutrons in @xmath1cl states having strong 1d@xmath37 or 2s@xmath40 neutron configurations would tend to populate states in @xmath21cl of sufficient excitation energy to emit a detectable , @xmath42 3 mev , @xmath10-ray .
thus , about 8 neutrons per @xmath1cl nucleus , or 1.4x10@xmath44 neutrons in the @xmath1cl , ( the product n@xmath13 @xmath14 in eq .
1 ) would have appropriate binding energy to decay to states in @xmath21cl that could produce a @xmath10-ray above 3 mev in energy , so would have a high probability of being detected . if the detection efficiency for the @xmath46 is assumed to be 70% , and the lifetime for this decay process is 10@xmath45 years , one would expect to observe 3 - 4 events per year from neutron decay in @xmath1cl . although this decay mode would not produce as strong a limit on nucleon decay as would the lead , its signature would be considerably more definitive than that from the lead , because its backgrounds are so much more readily identifiable .
the primary background for decay of a neutron in @xmath1cl comes from neutrinos produced in the earth s atmosphere , but such events would not be able to simulate nucleon decay events .
although once the neutron was knocked out of the @xmath1cl , the resulting @xmath48 coincidence would be the same as for the neutron decay event , the atmospheric neutrino would emit a neutron , which would be detected to give its characteristic 8.6 mev @xmath10-ray .
this is a considerably greater energy than could be produced by any of the @xmath10-rays resulting from the de - excitations in @xmath21cl ; thus that neutron could be used to veto this type of background event .
the limit that can be achieved from any process such as the decays described above clearly depends on the detection efficiency .
thus we have run simulations of the events from decays of @xmath1cl based on the geant @xcite detector simulation software .
as planned , the lpc detector will consist of multiple independent cylindrical modules having radii of 2 m and heights , which are adjustable in the simulations , ranging from 1 to 2 m. the geant simulations were based on a single module .
the interior of the cylinder was assumed to be viewed by photomultiplier tubes ( pmts ) at both ends .
current design plans are to use an lpc solution of concentration , also adjustable in the simulations , of 75% to 80% lpc by weight . from the decay of a neutron in @xmath1cl ,
geant generated both the @xmath10-rays from the de - excitation of the @xmath21cl nucleus and the resultant @xmath47 from the decay of the @xmath21cl .
the initial starting state for @xmath21cl was randomly chosen from the energy levels that were below the proton separation energy .
@xmath10-rays were then generated to simulate the de - excitation of the @xmath21cl nucleus from the selected starting state down to either the ground state or the isomer .
the number and energies of the generated @xmath10-rays were based on data on the levels and @xmath10-rays for the @xmath21cl nucleus .
events that decayed to the isomeric state were rejected .
the fraction of the events that de - excite down to the ground state was found to be @xmath49% .
the @xmath21cl decay time is chosen by randomly selecting whether a decay will occur during the 1 ms time unit , and was repeated until a decay occurred .
the chance for a decay within a time unit was based on the half - life of @xmath21cl . if the decay time took longer than 40 s , the total accumulated time was set to zero and the decay test continued .
the kinematics for the @xmath47 were generated after the @xmath10-rays were generated and tracked .
the initial energy of the @xmath47 was selected from a lookup table and the decay time was added to the initial time of flight .
a gaussian profile was used for the initial energy distribution of the positron .
the parameters for the gaussian were derived from a fit to the @xmath47 energy spectrum calculated using the radlst program @xcite . upon entering a pmt , a photon s energy and time of flight were stored . as @xmath47s and @xmath50s have a large rms multiple scattering in lpc ,
any patterns in the cherenkov radiation are destroyed .
however , an event can be identified by the number of pmts that have fired in a localized area within a specified time window .
the requirement imposed to identify @xmath10-rays and @xmath47s required three pmts to fire in a localized cluster within a 20 ns window . to estimate the efficiency for detecting neutron decay in the @xmath1cl in lpc , 10,000 events were generated in a 100 cm deep tank with an 80% solution of lpc by weight .
the attenuation length of the lpc was taken to be 4.2 m @xcite .
the mean efficiency for detecting the @xmath10-rays from the de - excitation of the @xmath21cl nucleus over the volume of the detector was found to be @xmath51% . for detecting the @xmath47s ,
the mean efficiency was found to be @xmath52% , making the mean efficiency for detecting a coincidence between the @xmath10s and the @xmath46s to be @xmath53% . for a 200 cm deep tank with an 80% lpc solution , the mean detection efficiency for @xmath10s , @xmath47s , and coincidences
was found to be @xmath54% , @xmath55% , and @xmath56% respectively .
thus the efficiencies assumed above are consistent with those determined from the monte - carlo simulations .
for the 200 cm deep tank , the neutron detection efficiency was found to be @xmath57% .
the position of a nucleon decay event can be deduced from the time difference @xmath58 t between hits at the two ends of the tank . @xmath58
t can be defined in a variety of ways , _
e.g. _ by taking it to be the time difference between the peaks of the signals , either from the @xmath10s or the @xmath47s , from the two sides of the module , or by attempting to average early hits to utilize the leading edges of the signals on the two sides to do the timing .
the first approach can be confused somewhat by the arrival of photons scattered from the opposite side , whereas the latter can be complicated by limited statistics .
either approach seems to allow localization of events to about @xmath59 cm .
spatial localization is important , as it allows use of a position dependent neutron detection efficiency , so can allow greater emphasis on events that occur in the center of the detector , which will have a greater efficiency for detecting the veto neutron , than those at the edges , which will have a lesser efficiency .
given that the suggested searches for nucleon decay would be conducted in a supernova neutrino detector , these searches would have a long time , probably more than 20 years , to run .
thus the suggested experiments have the potential to improve greatly the existing limits on the n @xmath60 decay mode .
this work began at an aspen institute summer workshop .
it has continued under the support of us department of energy grant de - fg03 - 88er40397 , nsf grant phy-0099476 , and the swiss nsf ( grant 2000 - 061031.02 ) .
tr acknowledges support by a profil professorship by the swiss nsf ( grant 2024 - 067428.02 ) .
pati and a. salam , phys .
d * 8 * , 1240 ( 1973 ) ; phys .
* 31 * , 661 ( 1973 ) ; phys .
d * 10 * , 275 ( 1974 ) .
h. georgi and s.l .
glashow , phys .
rev . lett . * 32 * , 438 ( 1974 ) .
k.s . babu , j.c .
pati , and f. wilczek , nucl .
* b566 * , 33 ( 2000 ) .
babu and r.n .
mohapatra , phys . lett . *
b518 * , 269 ( 2001 ) .
t. appelquist , b.a .
dobrescu , e. ponton , and h .- u .
yee , phys .
letters * 87 * , 181802 ( 2001 ) .
h. hagiwara _ et al .
_ , particle data group , phys .
d * 66 * , 010001 ( 2002 ) ; http://pdg.lbl.gov y. totsuka , in _ proceedings of the 7th workshop on grand unification / icoban , 1986 _ , edited by j. arafune ( world scientific , singapore , 1986 ) , p. 118
h. ejiri , phys .
c * 48 * , 1442 ( 1993 )
. y. suzuki _ et al .
b311 * , 357 ( 1993 ) .
y. kamyshkov and e. kolbe , phys .
d * 67 * , 076007 ( 2003 ) .
j. pati , a. salam , and u. sarker , phys .
letters * b133 * , 330 ( 1983 ) .
mohapatra and a. perez - lorenzana , phys .
d * 67 * , 075015 ( 2003 ) . c. berger _
et al . _ ,
letters * b269 * , 227 ( 1991 ) .
boyd , a. stj .
murphy , and r.l .
talaga , nucl . phys . *
a718 * , 222c ( 2003 ) .
elliott , phys .
c * 62 * , 065802 ( 2000 ) . evaluated nuclear structure data file , http://www.nndc.bnl.gov/nndc/ensdf t. rauscher and f .- k .
thielemann , at .
data nucl .
data tables * 75 * , 1 ( 2000 ) t. rauscher amd f .- k .
thielemann , at .
data nucl .
data tables * 79 * , 47 ( 2001 ) e. kolbe , k. langanke , s. krewald , and f .- k .
thielemann , nucl . phys . *
a540 * , 599 ( 1992 ) t. rauscher , f .- k .
thielemann , and k .- l .
kratz , phys . rev .
* c56 * , 1613 ( 1997 ) a. bohr and b.r .
mottelson , _ nuclear structure _ , ( world scientific , singapore,1998 ) , p. 239
geant 3.21 , cern program library , long writeup w5013 ( 1993 ) .
burrows , computer code radlst , brookhaven national laboratory report no .
bnl - ncs-52142 , 1988 ( unpublished ) . | lead perchlorate , part of the omnis supernova neutrino detector , contains two nuclei , @xmath0pb and @xmath1cl , that might be used to study nucleon decay .
both would produce signatures that will make them especially useful for studying less - well - studied neutron decay modes , _
e.g. _ , those in which only neutrinos are emitted . |
polycrystalline samples of tripod kagome lattice ( tkl ) compounds mg@xmath0re@xmath1sb@xmath1o@xmath2 ( re = gd , dy , er ) were synthesized by solid state reactions .
stoichiometric ratios of re@xmath0o@xmath1 ( re = gd , dy , er ) , mgo , and sb@xmath0o@xmath1 powder were carefully ground and reacted at a temperature of 1573 k for 60 hours with several intermediate grindings .
performing the reaction at temperatures above 1623 k will introduce site - disorder between the re site and the mg site , which is evidenced by the cubic pyrochlore phase in x - ray diffraction ( xrd ) patterns .
the room temperature xrd patterns were measured with a huber x - ray powder diffractometer with the structural refinements performed using software package _ fullprof - suite_. the dc susceptibility measurements were performed using a commercial superconducting interference device ( squid ) magnetometer with a magnetic field of 200 oe . the ac susceptibility was measured at national high magnetic field laboratory with the conventional mutual inductance technique at frequencies between 80 hz and 700 hz .
the low temperature specific heat measurements were performed in a he3-he4 dilution refrigerator using the semi - adiabatic heat pulse technique .
the powder samples were cold - sintered with ag powder , the contribution of which was measured separately and subtracted from the data .
for all the specific heat data shown below , the magnetic contribution ( c@xmath62 ) was obtained by subtracting a lattice contribution estimated from the results of a separate measurement of the non - magnetic isomorph zn@xmath0la@xmath1sb@xmath1o@xmath2 .
the x - ray diffraction ( xrd ) pattern and rietveld refinement result for mg@xmath0gd@xmath1sb@xmath1o@xmath2 are shown in fig.[fig : s1 ] . the xrd pattern for the other two compounds are similar to that of mg@xmath0gd@xmath1sb@xmath1o@xmath2 except for the peak shifts due to the lattice parameter differences .
as expected , the lattice parameters decrease with the decreasing ionic size from gd@xmath8 to er@xmath8 .
the refined crystallographic results of atomic positions , lattice parameters , and selective bonds lengths are listed in tab .
[ tab:1 ] .
[ tbp ] [ cols="^,^,^,^,^",options="header " , ]
the magnetization curves at 2 k for three tkl compounds and their pyrochlore cousins ( all in polycrystalline forms ) are shown shown in fig .
[ fig : s2 ] , where apparent similarities are found for all three sample sets . for gd - tkl
, the magnetization curve shows a straight line up to 3 t , which signatures a isotropic g - factor as expected for a s state of gd@xmath8 ( j = 7/2 , l = 0 ) .
the magnetization reaches 7.0 @xmath55/gd@xmath8 ( 6.4 @xmath55/gd@xmath8 for gd@xmath0ti@xmath0o@xmath7 ) at 6.5 t , which is about the 88.5% of its effective moment .
similar to dy@xmath0ti@xmath0o@xmath7 , strong anisotropic behaviors are observed in dy - tkl that the magnetization quickly saturates at a plateau of 5.1 @xmath55/ dy@xmath8 , which is about half of its effective moment . in the pyrochlore spin ice ,
such half magnetization plateau is a characteristic behavior of dy@xmath8 ising moment @xcite .
similar half magnetization plateau observed here provides a strong evidence for the local ising anisotrpy for dy ion in the tkl system . for er - tkl ,
the magnetization reaches 4.8 @xmath55/er@xmath8 ( 4.6 @xmath55/er@xmath8 for er@xmath0ti@xmath0o@xmath7 ) , which is about half of the effective moment per er ion .
moreover , the magnetization curve of er - tkl is also very similar to that of its pyrochlore cousin er@xmath0ti@xmath0o@xmath7 for the whole field region from 0 to 6.5 t. this similarity provides a supporting evidence for the local xy anisotropy that has been confirmed in the er pyrochlore .
for the kagome lattice described in the main text , @xmath37 and @xmath38 are basis vectors of the triangular bravais lattice where there are three basis sites in a unit cell for an upright triangle of basis spins , labeled as blue , red and green . the nearest distance between two spin are thus @xmath63 = @xmath64 .
the positions of the bravais lattice points are denoted by @xmath65 , whereas each of the three sites in the unit cell is labeled by @xmath66 ( where a = blue , red , green ) .
so each site can be labled as @xmath67 = @xmath65 + @xmath66 .
the general hamiltonian of the system is : @xmath68 @xmath69 the @xmath70 is the vector between two classical spins @xmath71 , @xmath72 of the unit length . @xmath10 and @xmath73 index the unit cell , @xmath74 and @xmath75 run over the sites in the unit cell , while @xmath41 and @xmath42 denote the x , y , z components of the spin vectors .
the first term in eq.(2 ) is the interacting term while the second term is the single ion term .
fourier transformation of the equation ( 2 ) yields the hamiltonian in the wave vector space that is listed in the main text .
@xmath76 represents the local ising axis of site @xmath77 as given in the main text . to enforce the spin anisotropies ( e.g ising - like for dy ion and local xy - like for er ion ) , the sign of coefficient @xmath78 is taken negative for dy ion and positive for er ion while the magnitude of @xmath78 is chosen ten times larger than any other energy scale .
@xmath78 is zero for the half filled spin case of gd .
the first term in @xmath79 represents dipolar interaction and @xmath44 is the dipolar energy scale that takes the form : @xmath80 where @xmath25 is the effective moment of re ions .
for each re - tkl , @xmath63 is found by the retvield refinement of the xrd pattern and listed in table.1 .
the second term in matrix @xmath79 represents the nearest neighbor exchange interaction .
the effective exchange constant @xmath45 is positive for antiferromagnetic exchange coupling and negative for ferromagnetic exchange coupling .
an estimation of the exchange constant @xmath46 can be found from the measured weiss temperature @xmath81 via a mean field theory @xcite : @xmath82 where @xmath83 is the number of nearest neighbor . in our tkl system ,
@xmath83 = 4 . since @xmath45 is the effective exchange constant that couples two unit spins , @xmath84 .
because of the large spin moment in the re - oxide systems , @xmath81 has the contribution from both the exchange and dipolar parts . as a simple approximation
, one can estimate the exchange contribution by subtracting @xmath81 by the dipolar contribution @xmath44 @xcite .
thus we get : @xmath85 adopting the parameters for our tkl system , we get ( 1 ) @xmath44 = 0.79 k , @xmath45 = 6.1 k for gd compound , ( 2 ) @xmath44 = 1.31 k , @xmath45 = 1.12 k for dy compound , and ( 3 ) @xmath44 = 0.11 k , @xmath45 = 11.0 k for er compound . in order to determine their ground states ,
we have used a luttinger - tisza type theory to calculate the eigenvalues and eigenfunctions of the interaction matrix , following two early references by luttinger - tisza @xcite and onsager @xcite .
these fundamental papers outline a procedure for simple cubic cell with a single magnetic species
. their idea is adapted for the case of kagome lattice , viewed as having three independent spins in a triangular unit cell .
we compute the 9@xmath439 interaction matrix of @xmath40 at each wave vector numerically , and search for the absolute minimum eigenvalue . by scanning all wave - vectors of the brillouin zone ( bz ) ,
typically the minima lie on high symmetry lines , we evaluate all possible periodicities of the magnetic state .
the eigenvectors are decomposed to find the magnitude of moments on each of the three spins , if these are normalizable to the same constant we have an acceptable solution .
the case of a degenerate eigenvalue is subtle and often comes up in our analysis . in this case
, we perform a test : if no linear combination of the eigenvectors can be found leading to an equal normalization of the three spins , then the solution is unphysical and must be discarded . whenever the minimizing wave vector is different from @xmath86 or m point of the bz , the solution found by this method is analogous to a mean field spin density wave .
since the magnitudes of the spins , which is given by the real part of the wave , must vary in magnitude at different sites , this does not fulfill the conditions of luttinger and tisza .
for all the tkl system , the magnetic specific heat ( c@xmath62 ) was obtained by subtracting the total specific heat ( c ) by a lattice contribution estimated from the results of a separate measurement of the non - magnetic isomorph zn@xmath0la@xmath1sb@xmath1o@xmath2 .
below 6 k , the lattice contribution is almost two magnitude smaller than the magnetic contribution as shown in fig .
[ fig : s3](a , c , e ) . the corresponding integrated magnetic entropies ( s@xmath62 )
are shown in fig .
[ fig : s3](b , d , f ) . for mg@xmath0er@xmath1sb@xmath1o@xmath2 , the integrated s@xmath62 from 0.11 k to 6
k reaches 6.61 j / mol - er@xmath30k . due to the possible magnetic ordering at 80 mk and possible contribution from the @xmath87er nuclear spin , it is unlikely that the integrated entropy has any significance .
99 k. matsuhira , z. hiroi , t. tayama , s. takagi , and t. sakakibara , j. phys . : condens .
matter * 14 * , l560 ( 2002 ) . c. kittel , _ introduction to solid state physics _
, p446 , 7th ed .
( john wiley & son , inc . 1995 ) .
b. z. malkin , t. t. a. lummen , p. h. m. van loosdrecht , g. dhalenne , and a. r. zakirov , j. phys . :
condens . matter * 22 * , 276003 ( 2010 ) .
j. m. luttinger and l. tisza , phys . rev . * 70 * , 954 ( 1946 ) .
l. onsager , j. phys . chem . * 43 * , 189 ( 1939 ) . | we present the structural and magnetic properties of a new compound family , mg@xmath0re@xmath1sb@xmath1o@xmath2 ( re = gd , dy , er ) , with a hitherto unstudied frustrating lattice , the tripod kagome " structure .
susceptibility ( ac , dc ) and specific heat exhibit features that are understood within a simple luttinger - tisza type theory . for
re = gd , we found long ranged order ( lro ) at 1.65 k , which is consistent with a 120 @xmath3 structure , demonstrating the importance of diople interactions for this 2d heisenberg system . for re = dy , lro at 0.37 k is related to the
kagome spin ice ( ksi ) " physics for a 2d system .
this result shows that the tripod kagome structure accelerates the transition to lro predicted for the related pyrochlore systems . for re = er , two transitions , at 80 mk and 2.1 k are observed , suggesting the importance of quantum fluctuations for this putative xy system . _
introduction. _ the two - dimensional ( 2d ) kagome lattice magnet ( klm ) has been a favorite in the theoretical condensed matter community since the experimental work on scgo @xcite , due to the strong frustration associated with its network of corner - shared triangles .
many exotic states are predicted , such as the quantum spin liquid ( qsl ) state @xcite , the spin - orbital liquid state @xcite , the kagome spin ice ( ksi ) state @xcite , dipolar spin order @xcite , the kosterlitz - thouless ( kt ) phase transition @xcite , quantum order by disorder ( qobd ) @xcite , and nematicity and supernematicity @xcite .
the large variety of exotic states predicted for kagome spin systems lies in contrast to a paucity of experimental systems .
early efforts include the exploration of langasites re@xmath1ga@xmath4sio@xmath2@xcite , which possess a distorted kagome lattice .
recent attention has been paid to vesignieite bacu@xmath1v@xmath0o@xmath5(oh)@xmath0 @xcite and herbertsmithite zncu@xmath1(oh)@xmath6cl@xmath0 @xcite .
the later one shows intriguing signs of qsl behavior @xcite . from a materials standpoint , however , these two systems are limited by ( i ) known defect prone structures @xcite , and ( ii ) the inability to substitute facilely on the magnetic site ( e.g with non - heisenberg spins ) to realize states other than the qsl . clearly then
, finding new klm - containing compounds with spin - type variability is a challenge of the highest order .
intriguingly , a 2d klm is naturally contained in the frustrated 3d pyrochlore structure . in pyrochlores re@xmath0x@xmath0o@xmath7 ( re = rare earth element , x = transition metal element ) , both the re@xmath8 and x@xmath9 sublattices form alternating kagome and triangular layers along the [ 111 ] axis as a result of the corner - shared tetrahedrons ( fig .
1(a ) ) @xcite .
however , the strong inter - layer interaction enforces three - dimensionality .
an exception is found in studies of dy@xmath0ti@xmath0o@xmath7 in a [ 111 ] magnetic field , which polarizes the triangular layer spins , effectively decoupling the kagome planes , leading to a ksi state @xcite .
obviously , if one can remove the magnetic moment of the triangular layer in the pyrochlore lattice , a re@xmath8 kagome - only lattice might be realized , enabling the study of intrinsic kagome physics .
because of various spin and spin anisotropies of different re@xmath8 ions , exotic and rich magnetic properties should be immediately available via the complex interplay among the spin - orbital coupling , dipolar interaction , and exchange couplings . in pyrochlores , for example , this interplay leads to multi-@xmath10 ordering @xcite with multiple field induced transitions @xcite for heisenberg spins in gd@xmath0ti@xmath0o@xmath7 , the spin ice state @xcite for ising spins in ho@xmath0ti@xmath0o@xmath7 @xcite and dy@xmath0ti@xmath0o@xmath7 @xcite , and qobd physics in the xy spin system er@xmath0ti@xmath0o@xmath7 @xcite .
then , what will be the magnetic ground states in the re - based klms ? in this letter
, we have created such a klm - mg@xmath0re@xmath1sb@xmath1o@xmath2 based on partial ion substitution in the pyrochlore lattice . here
, the triangular layers in the pyrochlore structure are occupied by non - magnetic mg@xmath11 ions , leaving the re@xmath8-kagome layer well isolated from neighboring layers .
we studied three representative systems ( re = gd , dy , er ) by dc- , ac - susceptibility ( @xmath12 , @xmath13 ) , and specific heat ( c@xmath14 ) measurements .
we present a spin hamiltonian and show the fundamental differences in collective behavior between the 2d klms and their 3d pyrochore cousins . _
structure. _ sample synthesis method and measurement setups are described in the supporting material @xcite .
the mg@xmath0re@xmath1sb@xmath1o@xmath2 ( re = gd , dy , er ) structure is rhombohedral with r-3 m space group in hexagonal expression .
compared with the pyrochlore lattice , the triangular layers of both the re@xmath8 and sb@xmath15 sublattices in the klm structure are occupied by mg@xmath11 ( the mg@xmath11 site can also be replaced by co@xmath11 @xcite , mn@xmath11 @xcite and zn@xmath11 @xcite ) .
thus the chemical formula can also be written as ( mg@xmath16re@xmath17)@xmath0(mg@xmath16sb@xmath17)@xmath0o@xmath18 , which is a pyrochlore ( re@xmath0x@xmath0o@xmath7 ) structure with @xmath19 re@xmath8 and x@xmath9 ions substituted in an ordered manner ( fig .
1(b ) ) .
it is noteworthy that for the x - ray diffraction pattern of mg@xmath0re@xmath1sb@xmath1o@xmath2 @xcite , the strongest peak for pyrochlore at 2@xmath20 @xmath21 30@xmath3 disappears completely and splits into two peaks , providing evidence for the absence of mg - re or mg - sb site - disorder @xcite . as shown below
, the c@xmath14 peaks at their phase transitions are very sharp , further underscoring the high degree of site order in the kagome layers .
this good kagome layer separation is likely due to the large ion size difference between mg@xmath11 and re@xmath8 . in this structure ,
the nearest neighbor distance between the re@xmath8 ions within a kagome layer remains similar to that of its pyrochlore cousin , and the re@xmath8-kagome layers are well isolated from each other by the non - magnetic mg@xmath11 and sb@xmath15 layers .
take mg@xmath0gd@xmath1sb@xmath1o@xmath2 for example , the nearest gd - gd distance within a kagome layer ( 3.678 ) is similar to that in gd@xmath0ti@xmath0o@xmath18 ( 3.600 ) , and much smaller than that between different planes ( 6.162 ) .
since the dipole - dipole energy goes as 1/@xmath22 , this leads to inter - layer energies an order of magnitude smaller than intra - layer energies .
thus , the kagome lattice in mg@xmath0re@xmath1sb@xmath1o@xmath2 is seemingly free of structural defects . in re@xmath0x@xmath0o@xmath7 ,
one important structural feature is that each re@xmath8 ion is surrounded by eight oxygens ( fig .
1(a ) ) with two shorter re - o1 bonds lying along the local-[111 ] axis and six longer re - o2 bonds forming a puckered ring .
this feature defines the crystal electric field ( cef ) and the g - factor which determines the ionic anisotropy for the re@xmath8 spins . in mg@xmath0re@xmath1sb@xmath1o@xmath2 , this local oxygen coordination is largely preserved .
the re ion is still surrounded by eight oxygens with the two shortest re - o1 bonds remain lying along the local-[111 ] axis ( fig .
1(b ) ) .
the difference is that the longer six re - o bonds are divided into two sets : four longest re - o2 bonds and two intermediate re - o3 bonds @xcite . since the cef degeneracy has already been lifted by the pyrochlore - like anisotropy for an effective spin-1/2 system , the dominant anisotropy remains the one distinguishing the puckered ring from the local-[111 ] oxygens , making this in - plane anisotropy most likely irrelevant for the ground state degeneracy .
given the high degree of site order , the large difference in separation between intra - plane and inter - plane re ions , it is appropriate to consider this a well - formed kagome structure .
in addition , the cef - driven single - ion anisotropy , which is vestigial from the parent pyrochlore structure , defines directions for either the ising spins or the xy - spin normal vectors that are neither uniaxial nor uniplanar .
this particular situation of three distinct axes with specific inter - axes angles will be important for understanding ordered spin configurations , as we show below . given the uniqueness of this structure and the need to distinguish it from klms with undefined local anisotropy , we call this the _ tripod kagome lattice ( tkl ) " _ , inspired by a tripod " formed by three re@xmath8 and one mg@xmath11 ion ( fig .
1(c ) ) .
[ tp ] _ magnetic properties. _ for mg@xmath0gd@xmath1sb@xmath1o@xmath2 , a curie - weiss ( cw ) fit from 50 @xmath21 300 k of @xmath23 ( fig .
2(a ) ) yields a weiss temperature , @xmath24 = -7.35 k and an effective magnetic moment , @xmath25 = 7.91 @xmath26 .
the negative @xmath24 value is close to that of gd@xmath0ti@xmath0o@xmath7 ( @xmath24 = -11.7 k ) @xcite .
the value for @xmath26 is consistent with @xmath25 = 7.94 @xmath26 expected for gd@xmath8 ( @xmath27 ) . with measurement frequencies ranging from 80 to 700 hz , @xmath13 shows a sharp and frequency independent peak at t@xmath28 = 1.65 k ( fig.2 ( b ) ) , indicating a lro transition .
this transition is further confirmed by a sharp peak at the same temperature in magnetic specific heat @xmath29 ( fig .
2(c ) ) .
the magnetic entropy between 0.2 and 6 k is 17.16 j / mol - gd@xmath30k @xcite .
this value is close to rln(2s+1 ) = 17.29 j / mol@xmath30k for a s = 7/2 , indicating a complete lro among the gd@xmath8 spins . for mg@xmath0dy@xmath1sb@xmath1o@xmath2 ,
the cw fit below 10 k yields @xmath24 = -0.18 k and @xmath25 = 10.20 @xmath26 ( fig .
2(d ) ) , consistent with the free ion moment of 10.63 @xmath26 for dy@xmath8 ( @xmath31 ) . in dy@xmath0ti@xmath0o@xmath7 @xcite ,
the small negative @xmath24 ( -0.20 k ) is due to competition between the dipolar interaction and super - exchange couplings of dy ions .
here , the similarity in local structure translates into similar - size coupling to the pyrochlore case , since the total spin - spin coupling is dominated by the dipolar interaction . with the ferromagnetic dipolar interaction ,
the negative @xmath24 again shows the antiferromagnetic(afm ) nature of the dy@xmath8-dy@xmath8 exchange interactions in mg@xmath0dy@xmath1sb@xmath1o@xmath2 . a transition to lro at @xmath32 = 0.37 k
is observed both in the @xmath13 ( fig .
2(e ) ) and @xmath29 ( fig .
2(f ) ) .
the integrated magnetic entropy below 6 k is 5.38 j / mol - dy@xmath30k @xcite , which is close to rln2 = 5.76 j / mol - dy@xmath30k , as expected for a kramers - doublet .
this suggests the dy@xmath8 spins fully order below 0.37 k. for mg@xmath0er@xmath1sb@xmath1o@xmath2 , the cw fit above 50 k yields @xmath24 = -14.25 k , and @xmath25 = 9.45 @xmath26 ( fig .
2(g ) ) , consistent with the free ion moment of @xmath33 = 9.58 @xmath26 for er@xmath8 ( @xmath34 ) .
the value for @xmath24 is close to that of the pyrochlore er@xmath35ti@xmath35o@xmath18 ( @xmath24 = -15.93 k @xcite ) .
the @xmath13 was measured down to 30 mk with a broad peak observed around 80 mk ( fig .
2 ( i ) ) , which shows weak frequency dependence .
the @xmath29 was measured down to 120 mk and exhibits a weak and broad peak around 2 k ( fig .
2(i ) ) . at this temperature ,
no anomaly is observed in @xmath13 , while an extremely weak anomaly ( 2 * 10@xmath36 emu / mole - er ) was seen in @xmath12 at 2.1 k that is perhaps related to the weak @xmath29 peak . _
theoretical investigation and discussion. _ the three systems discussed here - gd , dy , and er , are likely representatives of the three different spin types , heisenberg , ising and xy , respectively , evidenced by similar low temperature magnetization curves compared with their pyrochlore cousins @xcite . in the 3d pyrochlore systems discussed above , each spin type yields significantly different behavior . to uncover the possible differences among the spin types in the tkls ,
we have used a luttinger - tisza type theory @xcite and studied the eigenvalues and eigenfunctions of the interaction matrix in wave vector space .
we construct a 2d kagome lattice with @xmath37 and @xmath38 as basis vectors of the triangular bravais lattice where there are three basis sites in a unit cell for an upright triangle of basis spins , labeled as blue , red and green ( fig .
2(j ) ) .
thus the general hamiltonian for the tkl can be written as @xcite : @xmath39 where @xmath40 is the sum of a dipolar part , exchange part and a single ion anisotropy part dictated by the cef effects . here
@xmath41 and @xmath42 are the cartesian indices of the spins and a , b run over the three basis sites in unit - cell .
the spin vector is the fourier component of the real space object , and @xmath10 runs over the brillouin zone of the triangular lattice .
thus for a given value of @xmath10 , @xmath40 is a 9@xmath439 matrix that can be easily diagonalized .
the dipolar part ( @xmath44 ) is fixed exactly by the effective moment of spin and distances between re ions , while the exchange ( @xmath45 ) and single ion terms are found from the @xmath24 and the cef splitting of the re@xmath8 in the given environment @xcite .
the mg@xmath0gd@xmath1sb@xmath1o@xmath2 has heisenberg spins with @xmath46 = 7/2 , @xmath47 = 0 and therefore no single ion term .
the @xmath24 of -7.35 k leads to an estimate of the exchange constant @xmath45 @xmath21 6.10 k , while the dipolar energy scale of nearest neighbor spins @xmath44 @xmath21 0.79 k. we found that the minimum eigenvalue of @xmath40 is at the brillouin zone(bz ) center with @xmath10 = ( 0 , 0 ) , and the corresponding eigenvector represents a 120@xmath3 state where the three spins in the unit cell lie in the plane pointing along three axes at angles 2@xmath483 to each other ( fig.2 ( j ) ) . here
, the large dipolar term breaks the rotation invariance , lifts the frustration of a kagome lattice and helps defeat the mermin - wagner theorem that forbids magnetic ordering in a 2d heisenberg lattice .
it is known that higher values of spin than 1/2 releases the frustration somewhat like soft spins would @xcite , and the case here has s = 7/2 .
this seems to enable a 2d - ising like transition with a logarithmic heat capacity la onsager .
actually , similar spin structure was predicted by maksymenko et .
al by considering classical dipoles on a kagome lattice@xcite .
their calculated specific heat actually agrees well with our experiment in the critical region by proper scaling ( fig .
2(c ) ) .
thus , we conclude mg@xmath0gd@xmath1sb@xmath1o@xmath2 to be a rare example of dipolar interaction mandated spin ordering on a kagome lattice .
the dy@xmath8 ion is an effective spin-1/2 kramers ion with ising anisotropy in an eight - oxygen - surrounding environment .
similar to the spin ice system , the ising axis in dy - tkl variant is along the lines joining each dy to o1 ( dashed lines in fig .
1(c ) ) . for the three sites in our bravais lattice ,
the ising directions are @xmath49 = @xmath50 , @xmath51 = @xmath52 , @xmath53 = @xmath54 in the global cartesian frame .
the small @xmath24 of -0.18 k corresponds to @xmath45 @xmath21 1.12 k , while the dipolar energy scale @xmath44 @xmath21 1.31 k is the largest energy scale .
it is known that ferromagnetic spins with tripod - like anisotropy on a kagome lattice are highly frustrated , which will lead to the ksi state @xcite .
similar to that of the pyrochlore spin ice @xcite , the ice rule ( spins with either two - in - one - out or one - in - two - out respect to the center of each triangular ) of ksi will also result in great number of ground state degeneracy and zero - point entropy @xcite . with the tkl and strong dipolar interaction , it is tempting to view mg@xmath0dy@xmath1sb@xmath1o@xmath2 as a realization of a dipolar ferromagnet where the ksi physics could be realized .
our luttinger - tisza method yields a metastable state at the bz center , which is an ordered ksi state . the corresponding spin structure ( fig .
2(k ) ) can be viewed as a three - sublattices ferromagnetic order with @xmath10 = 0 .
this spin structure also resembles the theoretically predicted lro state @xcite for the 3d pyrochlore spin ice observed in tb@xmath0sn@xmath0o@xmath7 @xcite .
however , this @xmath10 = 0 state is not a global ground state .
the lowest eigenvalue of the exchange matrix is found to be at the six k points of bz corners , whose energy is somewhat lower than that of the zone center .
such an eigenvalue corresponds to a lro state with a 3@xmath433 tripled magnetic unit cell .
in addition , the magnitude of the moment differs in space , as prescribed by a commensurate spin density wave ( sdw ) state . unlike the @xmath10 = 0 state where the ksi ice rule is preserved for every mg - dy tetrahedron ( grey triangular in fig .
2(k ) ) , here one out of six tetrahedrons violates the local ice rule .
note that the luttinger - tisza method used here , is more akin to the mean field theory when applied away from the zone center or the m point of the bz .
this is a nontrivial and complex problem ( see e.g. ref .
@xcite ) , and requires further theoretical investigation .
regardless the exact nature of ordering , our tkl system then appears to enable the spin dynamics to be much more efficient , as compared to the 3d dy@xmath0ti@xmath0o@xmath7 compound .
this interesting contrast therefore provides a strong impetus to the study of the underlying dynamics . along with the observed lro at 0.37 k , _
the dy - ising - tkl might provide a rare example exhibiting a lro state that breaks the ksi degeneracy_. the er@xmath8 ion in mg@xmath0er@xmath1sb@xmath1o@xmath2 has a large angular momentum j = 15/2 . at low temperatures
, it reduces to an effective spin-1/2 as a result of the kramers - doublet .
the high temperature @xmath24 @xmath21 -14.52 k implies a large exchange energy @xmath45 @xmath21 11.0 k , while the dipolar energy scale is @xmath44 @xmath21 0.11 k by assuming a moment of 3 @xmath55 ( similar to that in the er pyrochlore ) .
this implies that below @xmath21 10 k the spins are locked up into the state preferred by the exchange .
the single ion anisotropy term in this case gives rise to a local - xy model , where the er@xmath8 are energetically favorable to lie in the local xy - plane perpendicular to the ising axis discussed above . in the er - pyrochlore ,
such a xy model , will give rise to a u(1 ) degeneracy in the spin hamiltonian at the mean - field level that allows the er@xmath8 spins to rotate continuously in the xy plane @xcite .
in er - tkl , similar xy degeneracy is preserved for the exchange part of the hamiltonian .
however , an arbitrarily small long range dipolar interaction will break the degeneracy . by diagonalizing the interacting matrix , a lowest energy eigenvalue
is found at the bz center whose eigenvector , by a curious coincidence , corresponds to the coplanar model exactly same as that of the gd compound ( fig .
2(j ) ) .
regarding the experimental observations for the er - tkl , since the 2.1 k anomaly in @xmath56 is extremely weak in terms of the entropy under the peak , and @xmath12 shows a similarly weak anomaly , the order parameter might be one that still allows significant fluctuations below its @xmath57 , reminiscent of a kt transition @xcite where spin - vortices form and bind .
then the 80 mk transition shown on @xmath13 is likely related to the predicted 120@xmath3 coplanar afm ordering .
if so , such a low ordering temperature ( frustration index @xmath58 = @xmath59 @xmath21 180 ) suggests the importance of quantum spin fluctuations in terms of suppressing the ordering temperature and selecting the ordered state . the weak frequency dependence on @xmath13 around the peak might indicate an increasing spin - lattice relaxation time as temperature
is decreased .
the importance of thermal coupling between a coherent spin system and the lattice needs to be understood for both identifying and potentially using quantum materials @xcite .
future experiments including neutron scattering and muon spin spectroscopy will be useful to identify the nature of the two transitions at 80 mk and 2.1 k. _ summary. _ we discovered a new 2d rare earth tkl mg@xmath0re@xmath1sb@xmath1o@xmath2 by partially substituting the ions in the cubic pyrochlore lattice . our studies on three samples with re = gd , dy , er have already related their magnetism to various exotic states including the dipolar spin order , the ksi , and the kt transition . due to the large variability of the spin sets in the rare earth family and the possibility of tuning the lattice parameters via chemical pressures , other exotic physics might also be realized .
the future exploration of the whole tkl family members is expected to open a new field in condensed matter physics and materials science studies for coming years , such as the pyrochlore did during the last two decades .
the work of b.s.s . at ucsc was supported by the u.s .
department of energy ( doe ) , office of science , basic energy sciences ( bes ) under award no .
fg02 - 06er46319 .
k.l . and y.x.w .
thank the support of the national natural science foundation of china ( grant no .
11275012 ) .
a.p.r was supported by nsf - dmr 1534741 , and j.t .
was supported by nsf dge-1339067 .
the work at nhmfl is supported by nsf - dmr-1157490 and state of florida and the doe and by the additional funding from nhmfl user collaboration .
99 a. p. ramirez , g. p. espinosa , and a. s. cooper , phys .
rev .
lett . , * 64 * , 2070 ( 1990 ) .
p. lecheminant , b. bernu , c. lhuillier , l. pierre , and p. sindzingre , phys .
rev .
b * 56 * , 2521 ( 1997 ) .
x .- g .
wen , _ field theory of many - body systems : from the origin of sound to an origin of light and electrons _ , reissue ed .
( oxford university press , new york , 2007 ) . l. balents , nature * 464 * , 08917 ( 2010 ) .
r. schaffer , s. bhattacharjee , and y. b. kim , phys .
rev .
b , * 88 * , 174405 ( 2013 ) .
a. s. wills , r. ballou and c. lacroix , phys .
rev .
b , * 66 * , 144407 ( 2002 ) .
m. maksymenko , v. r. chandra , and r. moessner , phys .
rev .
b , * 91 * , 184407 ( 2015 ) .
y. zhao , w. li , b. xi , z. zhang , x. yan , s. j. ran , t. liu , and g. su , phys .
rev .
e , * 87 * , 032151 ( 2013 ) .
s. sachdev , phys .
rev .
b * 45 * , 12377 ( 1992 ) .
t. picot and d. poilblanc , phys .
rev .
b , * 91 * , 064415 ( 2015 ) .
p. bordet , i. gelard , k. marty , a. ibanez , j. robert , v. simonet , b. canals , r. ballou and p. lejay , j. phys . : condens .
matter * 18 * , 5147 ( 2006 ) .
h. d. zhou , b. w. vogt , j. a. janik , y .- j .
jo , l. balicas , y. qiu , j. r. d. copley , j. s. gardner , and c. r. wiebe , phys .
rev .
lett . * 99 * , 236401 , ( 2007 ) .
a. zorko , f. bert , p. mendels , p. bordet , p. lejay , and j. robert , phys .
rev .
lett . *
100 * , 147201 , ( 2008 ) .
y. okamoto , h. yoshida , and z. hiroi , j. phys .
. jpn . * 78 * , 033701 ( 2009 ) .
t. h. han , j. s. helton , s. chu , d. g. nocera , j. a. rodriguez - rivera , c. broholm and y. s. lee , nature phys .
* 492 * , 406 ( 2012 ) . s .-
h .
lee , h. kikuchi , y. qiu , b. lake , q. huang , k. habicht , and k. kiefer , nat . mater .
* 6 * 853 ( 2007 ) .
y. tabata , h. kadowaki , k. matsuhira , z. hiroi , n. aso , e. ressouche , and b. fak , phys .
rev .
lett .
* 97 * , 257205 , ( 2006 ) .
j. d. m. champion , a. s. wills , t. fennell , s. t. bramwell , j. s. gardner and m. a. green , phys .
rev .
b , * 64 * , 140407 ( 2001 ) . a. p. ramirez , b. s. shastry , a. hayashi , j. j. krajewski , d. a. huse and r. j. cava , phys .
rev .
lett . , * 89 * , 067202 ( 2002 ) . s. t. bramwell and m. j. p. gingras , science , * 294 * , 1495 ( 2001 ) .
m. j. harris , s. t. bramwell , d. f. mcmorrow , t. zeiske and k. w. godfrey , phys .
rev .
lett . , * 79 * , 2554 ( 1997 ) .
a. p. ramirez , a. hayashi , r. j. cava , r. siddharthan and b. s. shastry , nature , * 399 * , 333 ( 1999 ) .
l.savary , k. a. ross , b. d. gaulin , j. p. c. ruff and leon balents , phys .
rev .
lett . * 109 * , 167201 ( 2012 ) .
see supplemental material at [ url will be inserted by publisher ] for details of sample synthesis , measurement setups , x - ray diffraction pattern , crystallographic table with selective bond lengths , theoretical model , estimation methods for d@xmath60 and j@xmath61 , and magnetic entropy .
k. li , y. hu , y. wang , t. kamiyama , b. wang , z. li and j. lin , j. solid state chem . * 217 * , 80 ( 2014 ) . w. t. fu and d. j. w. ijdo , j. solid state chem .
* 213 * , 165 ( 2014 ) .
m. b. sanders , j. w. krizana , and r. j. cava , j. solid state chem . * 4 * , 541 ( 2016 ) .
s. t. bramwell , m. n. field , m. j. harris and i. p. parkin , j. phys .
: condens .
matter * 12 * , 483 ( 2000 ) .
j. m. luttinger and l. tisza , phys . rev . * 70 * , 954 ( 1946 ) .
l. onsager , j. phys . chem . * 43 * , 189 ( 1939 ) .
o. nagai , s. miyashita and t. horiguchi , phys .
rev .
b * 47 * , 202 ( 1993 ) .
r. siddharthan , b. s. shastry , and a. p. ramirez , phys .
rev .
b * 63 * , 184412 ( 2001 ) .
r. siddharthan , b. s. shastry , a. p. ramirez , a. hayashi , r. j. cava , and s. rosenkranz , phys .
rev .
lett . ,
* 83 * , 1854 ( 1999 ) .
r. g. melko , b. c. den hertog , and m. j. p. gingras , phys .
rev .
lett . , * 87 * , 067203 ( 2001 ) .
i. mirebeau , a. apetrei , j. rodriguez - carvajal , p. bonville , a. forget , d. colson , v. glazkov , j. p. sanchez , o. isnard , and e. suard , phys .
rev .
lett . , * 94 * , 246402 ( 2005 ) .
t. takagi , and m. mekata , j. phys .
soc .
jpn . ,
* 62 * , 3943 , ( 1993 ) .
t. takagi , and m. mekata , j. phys .
soc . jpn . ,
* 64 * , 4609 , ( 1995 ) .
j. m kosterlitz and d. j. thouless , j. phys .
c : solid state phys . * 6 * , 1181 ( 1973 )
. m. a. schmidt , d. m. silevitch , g. aeppli , and t. f. rosenbaum , proc .
natl .
acad .
sci .
usa * 111 * , 3689 ( 2014 ) . supplemental material |
ground - based gravitational wave ( gw ) detectors are traditionally classified into large baseline interferometers and resonant antennas . recently
, a new class of gw detectors has been proposed , based on huge masses kept at cryogenic temperature and called dual detectors @xcite-@xcite . at difference from previous massive cryogenic antennas ( such as weber bars ) , the dual system do not aim at reaching the best peak sensitivity in a narrow band around a resonance frequency , but it takes advantage of elastic forces to achieve a useful sensitivity in a wide frequency range . for this purpose , it is no longer equipped with resonant mechanical amplifiers , and it needs a very sensitive readout . the goal is realizing a detector which covers the acoustic frequency region ( 1 - 5 khz ) .
a dual gw detector exploits two oscillation modes of a mechanical system with a readout symmetric with respect to the center of mass .
do to the geometry , the responses to the readout force of the two modes must be summed , while the responses to the tidal force of the gw are subtracted . in the frequency region between the two resonance frequencies , the susceptibilities of the two modes
are in anti - phase , giving a reduced response to the readout force and an enhanced response to the gw @xcite ( back - action reduction ) .
such behavior has been recently verified and explored experimentally in ref .
@xcite , where the authors study the mechanical response to radiation pressure of a cm - size plano - convex mirror which is part of a high finesse optical cavity .
the internal modes of the mirror analyzed are in the @xmath0 hz frequency and @xmath1 kg mass range .
it is worth pointing out that the back - action reduction mechanism is also the basis for a proposed scheme of quantum non - demolition measurement .
two further important effects are considered in the design of dual : a ) a clever readout geometry allows to be blind to some low - frequency acoustic modes that would bring thermal and back - action noise peaks in the useful frequency range ( ` mode selectivity ' ) ; b ) a broad interrogation area enables to reduce the effect of local deformations due to thermal motion and readout force , i.e. , from a different point of view , the effect of high frequency modes ( wide area readout ) .
different configurations of dual have been proposed and studied : two nested spheres @xcite , where the relevant modes are the first quadrupolar mode of the inner and outer sphere ; two nested cylinders @xcite , again acting on the first quadrupolar mode of the nested bodies ; a single hollow cylinder @xcite , exploiting its first and second quadrupolar modes ; a symmetric set of cylinders @xcite , where the first dual mode is given by the elastic link between them and the second one by their first oscillation mode . for any configuration ,
the possibility of an optical readout is extremely useful for its potentiality to reach a detection sensitivity at the standard quantum limit@xcite , or even surpass it@xcite .
an optical scheme expressly conceived for the application in a dual detector is proposed and studied in ref .
the idea is folding a fabry - perot cavity in order to interrogate a large detector area with a beam bouncing several times between the sensing masses , before being reflected back on itself by the end mirror to close the cavity .
such a kind of folded fabry - perot ( ffp ) has never been implemented experimentally . an interesting possibility for boosting the sensitivity of massive detectors to short bursts
is suggested by vitali _
et al._@xcite .
they show that , in some conditions , periodic cycles of fast feedback cooling of the sensitive mechanical mode , followed by a measurement period lasting a fraction of the temperature recovery time , brings an improvement of the sensitivity by even an order of magnitude .
cooling of a mechanical oscillator by means of radiation pressure@xcite has been recently demonstrated on micro - resonators using both active techniques ( cold damping)@xcite , with feedback acting on the optical power impinging on the mirror , or passive schemes ( self cooling ) , exploiting the mechanical effect of red - detuned laser radiation @xcite .
the above results were obtained on effective oscillator masses of less than 1 g , and frequencies in the @xmath2 hz range .
a recent work @xcite describes the cooling of a gram - size mirror held by a cantilever flexure , with a resonance frequency of @xmath3100 hz . here
the cantilever damping is obtained through a so - called hybrid scheme , exploiting active feedback on the cavity length which actually modifies the radiation pressure exerted by detuned laser radiation .
the cooling effect can be attributed to the optical spring originated by radiation pressure , modified ( with an added reactive component ) by feedback .
a similar hybrid technique is used also by corbitt _
et al._@xcite to cool a suspended @xmath31 g mirror , with the difference that here the detuning is modified by feedback acting on the laser frequency instead of the cavity length .
however , the system explored in ref .
@xcite is very peculiar : the optical spring is so strong that it completely determine the oscillation frequency ( around @xmath31 khz ) and the mirror confinement is provided by the optical potential rather than elastic forces .
a completely different cooling technique has been recently applied to the @xmath4 khz fundamental mode of the 3 tons aluminum bar of the auriga detector@xcite . in that work
the feedback is electrical and is applied to the capacitor of the mechanical readout . starting from a cryostat temperature of 4.2 k ,
the authors obtain a record temperature of 0.17 mk . due to the innovative dual concept ,
on the one hand a preliminary experimental study on a small - scale prototype is required .
on the other hand , it is important to scale up the experiments on opto - mechanical effects from the microscopic to the macroscopic range of masses and in the khz frequencies , and test the schemes that should be applied to reach the quantum regime of detection . in this work we present the experimental characterization of a prototype of dual detector , composed of two oscillators with a mass of the order of the kg ,
whose distance is read by a high finesse optical cavity .
the mechanical response function is measured by exciting the oscillators through modulated radiation pressure .
we demonstrate two of the above described effects crucial for the dual detectors : a ) the reduction of the contribution of local susceptibility due to average over a large interrogation area .
such effect is measured on the photo - thermal response@xcite thanks to the first implementation of a folded - fabry - perot cavity@xcite ; b ) the back - action reduction due to negative interference between acoustic modes . moreover
, we obtain the active cooling of an oscillation mode through radiation pressure , on our device , which is orders of magnitude heavier than any previously demonstrated radiation - pressure cooled system .
the oscillators are made with two 135 mm wide , 30 mm high aluminum masses ( test masses ) , fixed to frames along their short edge by @xmath3 mm thick lateral membranes .
the first mass is 30 mm thick and 0.33 kg heavy ; the second one is 40 mm thick and 0.44 kg heavy . the two oscillators are carved from single aluminum blocks , whose external parts are kept together by invar spacers and represent the external frames of the structure ( see fig .
( [ foto ] ) ) .
two rows of 12.7 mm diameter mirrors form the ffp : five on one mass , including a flat , 130 ppm transmission input mirror and a 1 m radius end mirror angled by 18@xmath5 , and four flat mirrors on the opposite mass .
the distance between opposite mirror surfaces is @xmath6 mm .
the input mirror is positioned on the outer side of the 30 mm thick mass with respect to the other mirrors of the row , for easier alignment .
the ffp cavity length is @xmath7 mm . due to the non - normal reflections , for each longitudinal mode the s - polarization resonance frequency
is detuned by 6 mhz from the p - polarization resonance ( the mirror coating is optimized for normal incidence ) , with no appreciable linewidth difference . the prototype is placed on a cantilever mechanical suspension in a thermally stabilized vacuum chamber .
the mechanical properties of the two coupled oscillators have been characterized by positioning accelerometers on the two frames and analyzing the response to a global mechanical excitation . a comparison with a finite element method ( fem ) model allows us to identify the peaks corresponding to the translation and torsion modes of the two masses ( their shapes are shown in fig .
( [ spettro ] ) .
in particular , the translation modes are found at @xmath3800 hz ( for the heavier mass ) and @xmath31200 hz ( for the lighter mass ) .
the same fem analysis gives the effective masses of the translation modes , excited and read on a small circular area corresponding to the central mirrors .
the results are respectively 0.31 kg and 0.41 kg , not far from the physical masses .
the experimental set - up for the opto - mechanical characterization of the prototype is sketched in fig .
( [ apparato ] ) .
the light source is a cw nd : yag laser working at 1064 nm .
after a 40 db optical isolator , the laser radiation is split into two beams . on the first one , a resonant electro - optic modulator ( eom1 )
provides phase modulation at 13.3 mhz with a depth of about 1 rad used for the pound - drever - hall@xcite ( pdh ) detection scheme .
the light is then transmitted by a polarization maintaining optical fiber and a second optical isolator ( o.i.1 ) .
the second beam can be shifted in frequency by means of two acousto - optic modulators ( aom ) and modulated in amplitude by an eom ( eom2 ) followed by an optical isolator ( o.i.2 ) .
we use respectively the + 1 and -1 diffracted orders of the aoms , so that the total frequency displacement corresponds to the difference of the aom frequencies and can be tuned by several mhz around zero .
after o.i.2 , part of the second beam is detected by a photodiode ( pd2 ) for monitoring its amplitude modulation .
the two beams are combined with orthogonal polarizations in a polarizing beam - splitter and sent to the folded optical cavity .
a quarter - wave plate allows us to optimize the matching of the polarization to the cavity modes . the reflected first beam , on his back path ,
is deviated by the input polarizer of the o.i.1 and collected by a photodiode ( pd1 ) for the pdh detection .
this pdh signal is used for laser frequency locking , while the second beam is tuned across the resonance using the aoms .
even when both beams are resonant with respect to the respective polarization modes , their frequency splitting allows to eliminate any spurious interference and cross - talk . the displacement noise spectrum has been obtained directly from the pound - drever signal with the laser weakly locked on the cavity , and stopping the second beam .
the signal is calibrated by modulating the laser frequency through its internal piezoelectric crystal , with a depth and a modulation frequency smaller than the ffp cavity linewidth , and using a phase - sensitive detection on the pdh signal .
the laser piezoelectric crystal is itself calibrated by observing the sidebands at 13.3 mhz on a wide , slow enough scan .
the pdh calibration procedure , performed varying the modulation frequency , allows also to correct the data for the servo loop . in order to characterize the response of the ffp to variations of the intracavity power ,
the laser is weakly locked on the cavity using the first beam while the second beam is amplitude modulated at different frequencies .
the extraction of the cavity response is obtained again from the pdh signal . the signal from pd2 and pd1
are acquired simultaneously with a digital oscilloscope , and successively elaborated to infer the component synchronous with the modulation .
the depth of the intracavity power modulation is calculated from the signal of pd2 by taking into account the mode matching and the cavity coupling factor .
the latter is inferred from the depth of the dip in the reflected beam when scanning on the resonance , and it is consistent with the independent measurements of cavity finesse and input mirror transmission . a first important test concerns the optical properties of the ffp configuration , that had not been implemented before .
we could indeed obtain a mode - matching of 93% ( comparable with that obtained with a simple cavity ) and a finesse of 6000 ( linewidth 125 khz ) with an intracavity optical power of the order of 1 w , in agreement with a calculation based on the independently measured losses of the mirrors .
the cavity length noise spectrum evidences the mechanical modes directly coupled to the optical cavity ( i.e. , modes changing the intracavity optical path ) . among them , it is possible to recognize the lower order translation and torsion modes , previously identified in the mechanical characterization of the cavity ( fig .
( [ spettro ] ) ) .
these mechanical resonances emerge from a background given by the free - running laser frequency noise ( in the frequency region of interest , it is approximately @xmath8 hz@xmath9/hz where @xmath10 is the frequency expressed in hz ) . the vertical scale in fig .
( [ spettro ] ) is expressed in terms of spectral density of variations @xmath11 of the cavity optical length @xmath12 ( measured from the input mirror to the end mirror ) , which are directly inferred from the measured spectrum using the calibration procedure described in the previous section .
for an intuitive understanding , the system can be approximated by infinitely rigid masses linked with mass - less membranes to a still frame . with this scheme ,
the translation modes corresponds to pure translations of the masses on the cavity plane and perpendicularly to the mirror surfaces , as shown by arrows in fig .
( [ apparato ] ) . in this case ,
simple geometrical considerations ( illustrated in fig .
( [ spostamenti ] ) ) lead to the relation @xmath13 where @xmath14 is the ( small ) variation of the distance between the two masses , @xmath11 is the corresponding change in the cavity optical length , and @xmath15 is the incidence angle on the folding mirrors ( @xmath16 in our case ) .
the expression ( [ dtol ] ) , inverted , can be used to convert the measured displacement from @xmath11 to @xmath14 . in this way , from the measurement in fig .
( [ spettro ] ) we can directly deduce the displacement of ideal translation modes , while for other mechanical modes one can only see in the spectrum the corresponding fluctuations of the cavity length . in the approximated scheme with rigid masses ,
the relevant torsion modes correspond to rotations of the mass around an horizontal ( modes a , b in fig .
( [ spettro ] ) ) or a vertical axis ( mode d in the figure ) . in this approximation
, such modes do not change the cavity length at the first order in the angular motion .
therefore , the system should be completely blind to the torsion modes , that are excited and detected only thanks to deformations in the masses and frames .
( left ) or a photo - thermal deformation @xmath17 ( right ) . in the case of the folding mirror ,
the path change is ( considering the right panel ) @xmath18 , while for the end mirror it is simply @xmath17 .
the incidence angle ( @xmath16 in our setup ) is exaggerated for clearness . ]
the mechanical quality factors of the modes , measured from their spectral width , are respectively 1800 ( heavy mass translation mode at 795 hz ) , 850 ( light mass translation mode at 1190 hz ) , and 310 ( heavy mass torsion mode around a vertical axis , at 1166 hz ) .
the ratio between the areas of the two peaks corresponding to the first order translation mode of the two masses is 1.45 , in fair agreement with the value of 1.69 calculated by assuming a white driving force noise .
a modal effective temperature @xmath19 can be calculated from these areas , according to @xmath20 where @xmath21 is the displacement noise spectral density , @xmath22 is the boltzmann constant , @xmath23 is the modal mass and @xmath24 its eigenfrequency . using the values for the mass calculated by the fem analysis
, we obtain for the two modes effective temperatures of about 3300 k and 3900 k respectively , clearly showing that the effects of external , technical noise is about one order of magnitude larger than thermal noise .
we can infer from this measurement an horizontal displacement noise , at the level of the ffp frames , of about @xmath25 m@xmath9/hz .
this is compatible with seismic noise around 1 khz , filtered by the optical table and the ffp suspension . between the two masses , according to eq .
( [ dtol ] ) . in this way ,
the spectral density shown in the graph is reduced by a factor of @xmath26 . in the inset ,
fit with thermal noise as in eq .
( [ noise1 ] ) ( dashed line ) and external noise as in eq .
( [ noise2 ] ) ( solid line ) . ] the excess external noise is also confirmed by the spectral shape around 1.2 khz , where we find three nearly - degenerate modes : as already mentioned the highest frequency ( @xmath31190 hz ) belongs to the translation mode of the light mass , the intermediate frequency ( @xmath31166 hz ) is a torsion mode of the heavy mass , while the lowest frequency mode ( around @xmath31142 hz ) remains unidentified . in the case of thermal noise excitation
, the contributions of all these modes should add incoherently in the noise spectrum , that should read@xcite : @xmath27}\ ] ] where for each mode @xmath28 is the susceptibility , @xmath29 the mass , @xmath30 the eigenfrequency and @xmath31 the quality factor , and @xmath32 is the thermodynamic temperature . on the other hand , an external noise @xmath33 should excite them coherently , and therefore be filtered by a superposition of modes response functions , such as : @xmath34}\right|^2 s_x^{ext}\ ] ] where the effective masses do not necessarily correspond to the previous ones . we have fitted the experimental spectrum , in the mentioned region around 1.2 khz , with the expressions ( [ noise1 ] ) and ( [ noise2 ] ) .
we remark that the two fitting procedures uses the same number of free parameters , namely the set @xmath35 , three amplitudes , and the coefficient of the frequency noise background .
concerning the amplitude , we have chosen to fix the mass of the translation mode at its fem value , leaving the other two masses as free and multiplying by an overall factor that can be identified with @xmath33 of expression ( [ noise2 ] ) or with the temperature in expression ( [ noise1 ] ) .
the two fitting curves are reported in the inset of fig .
( [ spettro ] ) .
the fit with expression ( [ noise2 ] ) is clearly better .
a quantitative evaluation , underlining the contribution of the wings of the peaks where the difference between the two fitting functions is more evident , is obtained by calculating the quadratic sum of differences between the logarithm of the experimental data and of the corresponding theoretical values ( in other words , the @xmath36 in logarithmic scale ) .
this indicator is a factor of 1.7 higher when using expression ( [ noise1 ] ) , with respect to eq .
( [ noise2 ] ) .
since the number of free parameters is the same , the analysis gives a strong indication in favor of the model behind eq .
( [ noise2 ] ) , with a flat external noise spectrum .
of course , this is confirmed by the unrealistically high temperature ( or low mass ) that would be required by expression ( [ noise1 ] ) .
however , we remark here that , in the case of multiple overlapping peaks , there is not merely an increased spectral power due to extra - noise , but also a change in the shape of the spectrum that can be a stronger , calibration - independent indication . in fig .
( [ risposta ] ) we report the phase and amplitude response of the ffp to a modulation of the intracavity radiation intensity .
the measured signal is reported in terms of changes in the cavity length @xmath12 . to obtain the response ,
the signal is divided by the amplitude of the modulation in the intracavity radiation pressure , giving a susceptibility in m / n .
following the discussion in the previous section , the derived quantity has an intuitive interpretation only for the ideal translation modes of the test masses . in this case , the relation ( [ dtol ] ) allows to find the variations of the mass positions and , as we will see later , to extract values of the modal masses directly comparable with the fem model and/or the physical masses . ) . ]
the mechanical resonances emerge on top of a ( nearly ) constant background associated with photo - thermal effect , i.e. , to mirrors thermal expansion due to heating by the absorbed intracavity radiation @xcite .
this effect is more meaningfully expressed in terms of m / w , as indicated on the left vertical axis of fig .
( [ risposta ] ) .
as will be discussed later in this section the quantitative explanation of the photo - thermal background implies the verification of the wide area readout effect . the complete response , between 700 hz and 1250 hz
, is fitted to the coherent superposition of a constant , complex background @xmath37 and four mechanical resonance , using the following function : @xmath38}+c_{phth } \ , .\ ] ] the fitting procedure considers at the same time both quadratures of the response , minimizing their distance respectively from @xmath39 and @xmath40 .
the function obtained , expressed as amplitude and phase , is reported in fig .
( [ risposta ] ) and the fitting parameters are summarized in table [ tab1 ] .
the factor @xmath41 is fixed to @xmath42 ; as we will show later ( in eq .
( [ lmodeffp ] ) and the following discussion ) , this coefficient allows to attribute the modal masses @xmath29 to vibrations of the test masses , excited and read on the laser spots .
.parameters obtained by fitting the experimental response to a modulated intracavity power with expression ( [ fitfun ] ) . [ cols="<,^,^ " , ] for the translation modes ( at 795 hz and 1190 hz ) we infer effective masses of 0.2 kg and 0.34 kg respectively . while for the latter the agreement with the physical mass and
fem simulation is good , for the former it is only marginal .
the effective mass of the torsion mode at 1166 hz is about 1 kg , much more than the physical mass involved in the motion , showing that , as expected , the readout is nearly blind to this mode .
also this effect is particularly important for dual , where the readout topology is studied in order to minimize the effect of disturbing mechanical modes within the useful detection band @xcite . concerning the mechanical quality factors , the values obtained from the fit are respectively 1940 ( at 795 hz ) , 835 ( at 1191 hz ) , and 370 ( at 1171 hz ) , all in good agreement with the corresponding figures given by the displacement noise spectrum .
the fourth resonance ( around 1140 hz ) is barely visible .
it is particularly interesting the region around 1.2 khz , shown in detail in fig .
( [ rispostazoom ] ) , where a torsion and a translation mode are very close . here
, between the two peaks , the complete response falls below each single peak contribution ( shown with dashed lines in the figure ) , due to their interference .
this effect implies a cancellation of the back - action : the system is less sensitive to modulation of the intracavity power , and therefore also to laser amplitude fluctuations and radiation shot - noise , which are the physical origin of the back - action . on the other hand
, the effect of a force acting on both modes in the same direction would be amplified .
the same should happen between the two gw sensitive modes of a dual detector .
the demonstration of this effect on our prototype would be even clearer by exploiting the translation modes of the two masses , for which the reaction to a gw is more intuitive , but it is prevented by the photo - thermal background . ) , together with a complete fitting function ( red , solid line ) , the contributions of two peaks ( respectively green dotted and blue dashed - dotted lines ) and contribution of the photo - thermal background ( gray dashed line ) . ]
let us now consider the other crucial issue of dual : the wide area readout. in the case of optical readout , it can be implemented by the multi - spot configuration of the ffp .
we first summarize its working principle , in the simple scheme of @xmath43 spots on one surface and @xmath44 on the opposite one .
we suppose that the spots are far enough to neglect the overlap between local deformations , and the beam impinges with small incidence angle ( more detailed studies are reported in ref .
@xcite ) . in this case , if the first surface moves by @xmath14 , the change in the optical path is @xmath45 , with an increase by a factor of @xmath46 with respect to a simple cavity .
the radiation pressure noise , due to intracavity power fluctuations , acts at the same time on all the spots .
the resulting displacement noise power in the optical path is @xmath47 , where @xmath48 are the fluctuations in the position of a single spot .
it seems that the signal - to - noise remains unchanged with respect to a simple cavity , with both noise and signal power multiplied by @xmath49 .
however , we must consider that in a ffp the intracavity power is reduced , and therefore also its fluctuations . indeed , the intracavity power is @xmath50 where @xmath32 is the input mirror transmission , @xmath51 the other roundtrip losses , @xmath52 the input power . for an optimally coupled cavity ( the best configuration for a pdh detection ) , with @xmath53 , one has @xmath54 .
now , in a ffp , @xmath55 where @xmath56 are the losses in a single spot ( in the round - trip , each spot is touched twice ) .
therefore the intracavity power is reduced by a factor of @xmath46 and its power fluctuations by @xmath49 , as well as @xmath48 . as a consequence
, the signal - to - noise power ratio is improved by a factor @xmath49 .
the calculation of the brownian thermal noise is even simpler : we have uncorrelated noise fluctuations in the different spots , therefore @xmath57 ( @xmath58 are the fluctuations in the ffp optical path , @xmath59 the displacement thermal noise in a single spot ) and the signal - to - noise is improved by a factor of @xmath43 . in this work
the sensitivity is not enough to see the effects of local thermal noise and radiation pressure effects : the former is overwhelmed by laser frequency noise , the latter by photo - thermal background .
however , for what the wide area readout is concerned , such photo - thermal expansion behaves exactly like a deformation due to radiation pressure : phase and amplitude of the response are different , but the effects on the different spots sum up in the same way . as a further difference with respect to the described model
, we can not optimize the input mirror transmission .
however , we can measure the ratio between mechanical modes peaks and photo - thermal background in the response to modulated intracavity power .
thanks to the broad area readout , the peaks emerge from background stronger than in the case of a simple cavity .
in particular , for our configuration , if @xmath60 is the photo - thermal displacement of each single spot due to a change @xmath61 in the intracavity power , the change in the ffp optical length is @xmath62 this is illustrated in fig .
( [ spostamenti ] ) : on the end mirrors the change of optical length is @xmath17 , while on the seven folding mirrors it is @xmath63 . in the case of a simple cavity , summing the effects on its two mirrors , we would have @xmath64 . due to a translation mode of one ffp mass , with susceptibility at resonance @xmath65
, we have instead @xmath66 to be compared with the simple cavity expression @xmath67 equation ( [ lmodeffp ] ) is obtained by using eq .
( [ dtol ] ) and the mass displacement @xmath68 where @xmath69 due to the four bounces .
we finally derive @xmath70 for the ffp , and @xmath71 for a simple cavity .
we have measured the photo - thermal expansion in a simple fabry - perot cavity with the same mirrors used in the ffp and roughly the same beam size ( the cavity length is 0.2 m)@xcite . in the frequency range of interest ( around 1 khz ) the effect has a weak dependence on the frequency , and for each single mirror is @xmath72 m / w . using this value in the expression ( [ dlphthffp ] ) , we infer @xmath73 m / w , in good agreement with the value of @xmath74 m / w obtained from the fit of the experimental data ( reported with a dashed line in fig .
( [ rispostazoom ] ) ) .
concerning the peak - to - background ratio , taking for instance the first translation mode at 800 hz with @xmath75 , we calculate @xmath76 and @xmath77 , to be compared with the experimental value of 540 .
the data presented here can be considered as the first experimental verification of the wide area readout effect , for the particular case of multi - spot readout .
the experimental scheme for the radiation pressure cooling of a ffp translation mode is similar to the setup implemented for the measurement of the response to modulated intracavity power .
the measurement beam is weakly locked to a resonance of the optical cavity , and its pdh signal is used also for obtaining the spectrum of the cavity length fluctuations .
the calibration of the spectral measurement is performed with the procedure described in section ( [ expsetup ] ) , even if now we are just interested in a narrow frequency region around the peak at @xmath3800 hz .
the same pdh signal , amplified and integrated , is used as correction signal acting on the second beam for the active radiation pressure cooling .
we have experimented two different cooling schemes . in the first case ,
the second beam is tuned on resonance of its polarization mode , and we act on its intensity ( by means of eom2 ) .
this configuration implements a standard optical cold damping , like the one firstly demonstrated in ref .
@xcite and later , i.e. , in ref . @xcite . with respect to the mentioned works , here we have a close spatial matching between sensing and cooling beams . in the second scheme ,
the cooling beam is frequency shifted on the average by half linewidth from resonance , and the feedback acts on the beam detuning ( by means of one aom ) which actually controls the intracavity power .
this configuration is somehow original : its working principle is the same as in ref .
@xcite ( the so - called hybrid scheme ) , with the difference that here we react on the laser - to - cavity detuning by changing the laser frequency instead of the cavity length .
an important characteristics that distinguishes our experiment from the those described in refs .
@xcite and @xcite is the use of two separate laser beams for measuring the mass motion and for cooling . in this way , the measured signal ( i.e. , the detuning between the probe beam and the cavity ) is not directly modified by the cooling servo - loop , that only acts on the frequency of the second beam without directly modifying the probe frequency , nor the cavity length .
this procedure eliminate the necessity to correct the acquired data for the feedback : the probe beam can be considered as free - running and the pdh signal gives directly the cavity length fluctuations , thus allowing a clearer interpretation of the results . the physics behind active cooling is described in refs .
@xcite , @xcite and in several subsequent articles .
we summarize here its main features , also in order to clarify our experiment . in general ,
a signal proportional to the detuning is frequency filtered and transformed into a force acting on the oscillator .
the evolution equation of the oscillator position , in the fourier space , reads @xmath78 where g is a general complex , frequency - dependent gain function and @xmath79 is the noise force , that in the case of brownian thermal noise has spectral density@xcite @xmath80 a general , white force noise can be included by replacing the thermodynamic temperature @xmath32 with a noise temperature .
( [ cool1 ] ) gives @xmath81 with @xmath82re@xmath83 and @xmath84 .
taking a feedback proportional to the oscillator velocity , e.g. , @xmath85 , the effective eigenfrequency and quality factor do not depend on @xmath86 , and the equation of motion is the same as the one of a free oscillator , with modified damping ( @xmath87 ) and temperature . in particular , for @xmath88 the quality factor is increased and the effective temperature , obtained from the integral of the spectrum as in eq .
( [ teff ] ) , is increased as @xmath89 .
the same expressions for negative @xmath90 gives lowering quality factor and temperature ( cold damping ) .
if @xmath91 has a different dependence on @xmath86 , the parallelism between the modified evolution equation and a free oscillator is not straightforward . even in simple realistic schemes the energy equipartition does not hold and a system temperature is not well defined , as discussed , e.g. , in ref .
however , for high quality factor , most of the fluctuations are localized in a narrow spectral region around resonance and we can use a lorentzian approximation of the spectrum , so that @xmath92}\ ] ] where in @xmath93 and @xmath94 we can replace @xmath95 by its value at @xmath96 .
even a colored external noise can be included in the discussion , provided that it has a smooth spectrum in the region of the peak . in this case
an effective temperature can still be defined according to eq .
( [ teff ] ) , that gives @xmath97 while the effective eigenfrequency is shifted by the real part of the gain . in the adiabatic limit , for frequencies well below the cavity linewidth
, the force exerted by radiation pressure can be written as @xmath98 where @xmath99 is the intracavity power at resonance and @xmath100 the detuning normalized to the cavity half - linewidth @xmath101 . in the standard cold damping scheme ,
the laser is kept at resonance and a signal proportional to the detuning is sent to correct the laser power . neglecting the laser frequency fluctuations and the cavity length noise ( except for the oscillating mirror position @xmath102 ) , the loop gain can be written as @xmath103 where @xmath104 is a ( complex and frequency - dependent ) electronic servo loop gain expressed as the ratio between the detuning fluctuations @xmath105 and the consequent imposed relative power fluctuations . in the hybrid configuration , the cooling laser is detuned and the feedback is on the detuning .
we can expand eq .
( [ frp ] ) around the working point @xmath106 , obtaining for the radiation pressure force fluctuations @xmath107 @xmath108 where @xmath109 is the average cavity length , @xmath110 its fluctuations , @xmath111 is the cooling laser frequency with fluctuations @xmath112 , and we have neglected laser amplitude noise .
the cavity and/or the laser frequency fluctuations may contain a term proportional to @xmath105 through electronic servo loop gains that we call respectively @xmath113 and @xmath114 .
this term is obtained from a measurement of the detuning between the cavity and a probe beam .
( [ frp ] ) can now be written in an interesting form as @xmath115\ ] ] where @xmath116 is the optical spring rigidity @xcite and @xmath117 is a general extra - noise term .
the meaning of @xmath118 is clear when replacing eq .
( [ frp2 ] ) in the expression for @xmath93 : the effect of radiation pressure is modifying the oscillator spring rigidity @xmath119 according to @xmath120 .
an oscillator damping is imposed by the reactive component of the optical spring . in the passive self - cooling
it is given by the delay in the optical field buildup inside the cavity ( that we have neglected in the adiabatic approximation ) .
in the active scheme , it is obtained by the feedback that , as we see in eq .
( [ frp2 ] ) , modifies the optical spring rigidity .
concerning the extra - noise , it includes a ) the cavity length noise @xmath121 entering both directly in @xmath110 and from the measurement exploited for cooling ( i.e. , through the servo - loop ) ; b ) the cooling laser frequency noise @xmath122 ; c ) the probe beam frequency noise @xmath123 .
we have therefore @xmath124 \ , .\ ] ] the right end side of eq .
( [ cool1 ] ) becomes @xmath125 and we are in the conditions of the previous general discussion by replacing @xmath126 and adding to the force noise spectrum @xmath127 the additional term @xmath128 , where @xmath129 is the spectral density of @xmath117 . in our case ,
the dominant term in @xmath129 comes from the probe beam frequency noise @xmath130 and the feedback on the cooling beam frequency is characterized by @xmath131 .
therefore , we can write @xmath132 and , expressing the gain as @xmath133 , the equation for the effective temperature given in eq .
( [ teff ] ) can be written in the useful form @xmath134 where @xmath52 is the input power of the cooling beam , the normalization constant @xmath135 is @xmath136 and the extra - noise coefficient @xmath137 is @xmath138 where @xmath139 is the displacement noise at resonance measured before cooling .
we remark that , for the optimal feedback phase @xmath140 , the value of @xmath137 can be inferred directly from the displacement noise spectrum : @xmath141 is the peak value , and @xmath142 is the background .
the minimum effective temperature @xmath143 , achievable for @xmath144 , is @xmath145 . and initial temperature @xmath32 ( symbols ) as a function of the cooling laser power , fitted by eq .
( [ teff2 ] ) ( red line ) . ]
we could achieve cooling of the translation mode both with cold damping and with the hybrid scheme .
an example of the displacement noise spectra modified by cooling is reported in fig .
( [ cooling ] ) for the second configuration ( with frequency feedback on the cooling beam ) .
here we focus on the translation mode of the heavy mass , around 800 hz , and we have transformed the cavity length fluctuations into mass position fluctuations using the relation ( [ dtol ] ) .
the effective temperature of the mode , estimated from the area of the spectral peak , is reduced by a factor of more than 10 , as shown in the inset of the figure .
the cooling factor is limited the laser frequency noise and , with strong cooling power , the peak of the mechanical resonance disappears completely in the frequency noise background .
the measured effective temperature is fitted by eq .
( [ teff2 ] ) fixing the extra - noise factor @xmath137 at the value of @xmath146 , measured form the spectrum shown in fig .
( [ spettro ] ) . for the highest cooling power level ,
a weak instability in the electronic feedback loop slightly enhances the background noise . as a consequence ,
the uncertainty in the last temperature measurement is higher and we have not included in the fit , that gives @xmath147 mw and therefore a best obtainable cooling of @xmath148 for @xmath149 mw .
we have presented an experiment aimed at demonstrating and testing the main new concepts on which are based the proposals of dual gravitational wave detector . in particular , we show the back - action reduction and the wide area readout effects .
our prototype also contains the first implementation of folded fabry - perot cavity , an optical scheme particularly conceived for dual .
we report a complete characterization of the mechanical response of @xmath3 kg mass oscillators , performed by exciting the system with radiation pressure force .
moreover , we describe active optical cooling of an oscillator mode , also exploiting an original scheme with feedback on the laser frequency .
such experiences scale up the mass by several orders of magnitude with respect to previous experiments on opto - mechanical effects , and focus on the @xmath3khz frequency region , of interest for large mass gravitational wave detectors .
therefore , they represent a significant step for realistic planning of the next generation of massive detectors , whose possible optical readout should be pushed to the quantum regime where radiation pressure effects are critical@xcite .
this work was partially funded by the european union ilias project ( no .
rii3-ct-2003 - 506222 ) .
we thank m. cerdonio and m. de rosa for useful discussions . | the next generation of gravitational wave ( gw ) detectors is expected to fully enter into the quantum regime of force and displacement detection . with this aim , it is important to scale up the experiments on opto - mechanical effects from the microscopic regime to large mass systems and test the schemes that should be applied to reach the quantum regime of detection . in this work we present the experimental characterization of a prototype of massive gw detector , composed of two oscillators with a mass of the order of the kg ,
whose distance is read by a high finesse optical cavity .
the mechanical response function is measured by exciting the oscillators though modulated radiation pressure .
we demonstrate two effects crucial for the next generation of massive , cryogenic gw detectors ( dual detectors ) : a ) the reduction of the contribution of local susceptibility thanks to an average over a large interrogation area .
such effect is measured on the photo - thermal response thanks to the first implementation of a folded - fabry - perot cavity ; b ) the back - action reduction due to negative interference between acoustic modes . moreover
, we obtain the active cooling of an oscillation mode through radiation pressure , on the described mechanical device which is several orders of magnitude heavier than previously demonstrated radiation - pressure cooled systems . |
cooling in lattice field theory is a technique for exposing the topological features of field configurations @xcite .
recently , using this method , evidence has been presented @xcite that instantons play a dominant role in determining hadron properties in quantum chromodynamics ( qcd ) .
this conclusion was based on the calculation of long - distance properties of a variety of hadronic correlation functions .
it has been argued , however , that the persistence of long - distance effects ( particularly , confinement @xcite ) is an inevitable consequence of the local nature of the cooling procedure and is not indicative of the underlying dynamics @xcite . to gain some more insight into what cooling is doing
we have extended our work @xcite on three - dimensional qed to four - dimensional qcd with su(2 ) color .
we first examine the behavior of the static quark - antiquark potential ( calculated from wilson loops ) and of the spin - dependent interaction ( calculated from chromo - magnetic field correlations ) under cooling . as in qed@xmath2
, there is a good indication that the confining behavior of the static potential persists even after a significant amount of cooling although the overall scale of the potential is greatly reduced . on the other hand ,
the spin - dependent potentials ( which are short - ranged ) disappear rapidly upon cooling . in the spin - spin interaction , a residual effect , which one may associate with instantons , of only a few percent of the orginal potential
was observed . next chiral symmetry breaking was calculated using staggered fermions in quenched approximation .
the chiral symmetry order parameter @xmath3 was calculated at a few nonzero quark mass values and compared to the result obtained for free fermions .
our first observation is that @xmath3 relaxes to its free - field value even after a small amount of cooling .
this is an indication that dynamical mass generation in the cooled configurations is small compared to that found in the uncooled vacuum .
the extrapolation to zero quark mass is problematic since , in principle , it requires an extrapolation to infinite volume first .
however , we can conservatively estimate that in the region of intermediate cooling ( 20 to 50 cooling steps ) the chiral condensate has been reduced by at least an order of magnitude .
finally , meson correlators for the spin-0 and spin-1 channels were calculated .
the behavior of these time - correlation functions upon cooling is in qualitative agreement with what would be anticipated given the behavior of the potentials and the chiral order parameter .
the meson correlators do not relax uniformly .
the relaxation rate is most rapid at small times and it also decreases as quark mass decreases . at short time separations
the spin-0 and spin-1 meson correlators reflect very clearly the effect of cooling on the spin dependent potential ( degeneracy of pseudoscalar and vector states ) and on chiral symmetry breaking ( parity doubling in the spin-1 channel ) . at larger times at fit to the pseudoscalar correlation functions yields `` mass '' values roughly 2/3 the size ( in lattice units ) of the uncooled masses .
section ii contains a description of the calculational methods used in this paper .
the results are presented in sec .
the usual plaquette action @xmath4 is used .
periodic boundary conditions were imposed on the gauge field links in all directions .
field configurations were generated using a heatbath monte carlo algorithm . for cooling ,
links were updated `` vectorially '' following the same checkerboard sequence as was used in the monte carlo algorithm .
each link in turn was replaced by a link proportional to the inverse of the sum of the `` staples '' of the plaquettes containing the link being updated so as to minimize the local contribution to the action .
some test runs were done with adiabatic cooling @xcite .
after 10 to 20 cooling steps the results become qualitatively the same as with full local minimization .
the physics of cooled configurations has been interpreted in terms of instantons .
we also monitor the topological charge as a function of cooling and , following chu _
@xcite , the simple transcription of @xmath5 to the lattice @xcite is used for the topological charge density @xmath6 where @xmath7 , @xmath8 are plaquettes of gauge field links .
the total topological charge is @xmath9 as is well known this simple definition of topological charge does not accurately represent the topological charge of the uncooled lattice configurations @xcite .
however , for sufficiently smooth ( cooled ) configurations one can see instantons with the expected continuum action .
the creutz ratio @xmath10 where @xmath11 denotes the @xmath12 by @xmath13 rectangular wilson loop , can be used to determine the string tension . for large loops , which obey the area law
, @xmath14 gives the string tension .
the potential @xmath15 between static quarks was also calculated directly by extrapolating wilson loops to large @xmath13 @xmath16 variance reduction methods @xcite were used in the computation of the wilson loops in the uncooled theory .
in addition to the confining central potential , spin - dependent interactions can also be calculated @xcite .
these are related to chromo - magnetic field correlations ( see ref .
@xcite for a simple derivation ) which are computed by making magnetic field insertions on @xmath17 wilson loops .
the spin - spin and tensor interactions are then given by @xmath18 and @xmath19 where @xmath20 and @xmath21 are expectation values of wilson loops with plaquette insertions ( at @xmath22 and @xmath23 ) corresponding to magnetic fields parallel and transverse to the spatial direction of the loop .
in practise , the magnetic field insertion @xmath24 that was used was the average over the eight spatial plaquettes whose corners lie on the wilson loop @xmath11 .
this corresponds to operator ii of ref .
@xcite .
in addition to observables constructed purely from gauge field variables we are also interested in how quarks behave in the cooled vacuum . in qcd a basic property of the vacuum is chiral symmetry breaking which can be studied most easily if staggered fermions are used .
the action for staggered fermions is @xmath25 + \sum_x \ , m \overline{\chi}(x )
\chi(x ) , \nonumber \\ & \equiv & \overline{\chi } \,\ , { \cal m}(\{u\ } ) \ , \chi , \label{sf}\end{aligned}\ ] ] where @xmath26 are single - component fermion fields , @xmath27 is the staggered fermion phase @xcite , @xmath28 is the mass in lattice units and the @xmath29 are gauge field links .
antiperiodic boundary conditions were used for the fermion fields in all directions .
the chiral symmetry order parameter is calculated from the inverse of the fermion matrix @xmath30 of eq .
( [ sf ] ) @xmath31 where v is the lattice volume and the angle brackets denote the gauge field configuration average . a random source method @xcite
was used to calculate @xmath32 .
sixteen gaussian random sources were used for each gauge field configuration .
meson correlation functions can be constructed from local bilinears of the single - component @xmath33 fields .
we consider two such correlators which after integration over the fermion fields , can be expressed in terms of the inverse of the fermion matrix as @xmath34^\dagger \right\},\ ] ] and @xmath35 \mbox{tr } \left\ { { \cal m}^{-1}(\vec x , t ; 0 ) \left [ { \cal m}^{-1}(\vec x , t ; 0 ) \right]^\dagger \right\}.\ ] ] these two functions describe the propagation of zero - momentum meson states of spin 0 and 1 respectively . as is well known , with local staggered fermion operators mixing between states of different parity can in principle occur . in practise ,
the spin-0 channel is essentially pure pseudoscalar and describes the pseudo - goldstone boson .
the spin-1 correlator is dominated by the vector meson ( at least in the uncooled theory ) with some admixture of axial - vector meson states .
most of the calculations were done on a @xmath36 lattice at @xmath37 .
this value was chosen as it is well into the scaling region for su(2 ) color .
for comparison some calculations were done at @xmath38 on a @xmath36 lattice and at @xmath37 on a @xmath39 lattice .
however , not all results will be shown here since they are qualitatively the same in all cases .
it is useful to consider first the topological properties of the gauge field configurations .
a sample of 300 configurations , separated by 100 heat - bath monte - carlo sweeeps after 4000 sweeps of thermalization , was analyzed . in the uncooled configurations our value for the ( lattice ) topological susceptibility @xmath40 of @xmath41 agrees well with the high statistics result of @xmath42 obtained by campostrini _
the simple transcription of the continuum topological charge operator is not a true topological quantity on the lattice @xcite .
it need not take integer values as can be seen in fig . 1 which shows a histogram of the number of configurations versus @xmath43 . however in cooled configurations which are sufficiently smooth , the operator @xmath43 does cluster around integer values .
this is shown in fig .
2 constructed from our sample of 300 configurations at 25 , 50 , 75 and 100 cooling steps .
it is also useful to examine the behavior of the action under cooling .
histograms of the average plaquette are plotted in fig . 3 for different amounts of cooling .
4 shows a scatter plot of @xmath43 versus average plaquette . at 100 cooling steps there is a fairly obvious instanton interpretation .
the vertical dashed lines in fig . 3d correspond to values of the total action of @xmath44 for n=1 , 2 , 3 and 4 .
after about 100 cooling steps the configurations are dominated by a single classical instanton .
75 cooling steps seems to be in a transition region @xcite .
single instanton peaks are seen in the action but there are many configurations which have a more complicated structure .
the region of 25 to 50 cooling has been interpreted as being dominated by multi - instanton anti - instanton fluctuations .
however one has to be aware that , as can be inferred from fig . 3 and 4 , these configurations are not simply superpositions of isolated ( noninteracting ) classical instantons .
the creutz ratio and static potential can be calculated from wilson loops as in eqs . ( 4 ) and ( 5 ) .
a @xmath36 lattice is too small to see the true asymptotic confining behavior but we can infer the general trend .
the creutz ratio @xmath14 calculated from our sample of 300 uncooled configurations is shown in fig .
5(a ) . on the same graph
we also show the creutz ratio after 25 cooling steps .
further cooling results in the creutz ratios of fig .
the corresponding results for the potentials are shown in fig .
the potentials are steadily reduced by cooling but the concave upward curvature is consistent with the idea put forward by teper @xcite that the string tension survives cooling , albeit , at increasingly large distances .
spin - dependent potentials are calculated from chromo - magnetic field correlations .
the spin - spin and tensor potentials are plotted in fig . 7 and 8 respectively .
the uncooled potentials are qualitatively consistent with the @xmath45 operator ii results of michael and rakow @xcite .
the spin - dependent potentials are short ranged and decrease rapidly upon cooling . in the region of 25 to 50
cooling steps the residual effect in these potentials is only at the level of a few percent of the uncooled values .
of course it has to be remembered that the comparison of cooled and uncooled results is being done here in terms of lattice ( not `` physical '' ) units .
it is clear that if one attempted to keep the magnitude of the potentials quantitatively similar in physical units while cooling a very large change in the lattice spacing would be required .
spontaneous chiral symmetry breaking is a basic property of qcd .
a signature for this phenomenon is the persistence of a nonzero value of the `` quark condensate '' in the limit of zero quark mass . to study chiral symmetry breaking we
calculate the expectation value of the local staggered fermion operator @xmath0 . for our @xmath36 calculation ,
40 configurations ( separated by 200 monte carlo sweeps ) at @xmath37 and staggered fermion masses @xmath46 and @xmath47 were used . fig .
9 shows @xmath0 versus quark mass ( in lattice units ) .
the results without cooling are consistent with those of billoire _
@xcite at the same value of @xmath48 .
the quantity @xmath0 was also calculated using 20 configurations on a @xmath39 lattice and quark masses down to @xmath49 .
the results are shown in fig . 10 . at the masses that are common to both calculations , the @xmath36 and @xmath39 lattice results are consistent .
the results for @xmath0 after 10 cooling steps and for massive free staggered fremions are also plottted in figs . 9 and 10 .
after 10 cooling steps the values of @xmath0 are already quite close to the free field results at the same value of @xmath50 .
therefore , dynamical mass generation seems to be quite small in the cooled vacuum . in principle
the extrapolation of @xmath0 to zero quark mass to extract the genuine chiral condensate requires that the infinite volume limit be taken first .
this would allow calculations at arbitrarily small quark masses .
in practise we can only do a limited number of calculations so the procedure of billoire _
@xcite is adopted .
three values of @xmath0 at the lowest nonzero mass are used to determine the coefficients of the expansion @xmath51 the extrapolated values of @xmath52 from this procedure are plotted in fig .
11 versus cooling up to 50 cooling steps .
unfortunately the values from the @xmath36 and @xmath39 lattice calculations do not agree indicating that perhaps the correct mass window is not being used for the extrapolation or that the extrapolation function is not adequate .
therefore only a qualitative conclusion is possible , namely , that the chiral condensate ( in lattice units ) is greatly reduced in cooled configurations .
of course , one could keep the condensate fixed in physical units but this would require a decrease of the lattice spacing by a factor of about 2.5 to 3 . finally we examine the meson correlation functions .
first we consider the qualitative effects of cooling by comparing the meson correlators in the vector and pseudoscalar channels . a sample of results at the lightest quark mass values for both @xmath36 and @xmath39 lattice are presented .
12 shows the pseudoscalar and vector zero - momentum correlators on the @xmath36 lattice for @xmath53 . after only a small amount of cooling the correlation functions for small time separations
display essentially the same behavior as would be obtained for free fields .
the pronounced oscillation of the vector correlator indicates near degeneracy of even and odd parity spin-1 states ( parity doubling ) .
it is clear that , after cooling , pseudoscalar and vector states are also nearly degenerate .
in other words , after 20 to 50 cooling steps , quark propagation at @xmath53 , at least in the limited time region available on the @xmath36 lattice , shows very little evidence of spin - dependent forces or chiral symmetry breaking .
on the @xmath39 lattice the correlations functions can be explored over a larger time range and at smaller masses .
the results for @xmath53 and @xmath54 are plotted in figs .
13 and 14 respectively . at large time separations
the correlation functions differ from those obtained for free fields .
for example , comparing figs . 12 and 13 one sees at the largest time separation some evidence of a difference between the pseudoscalar and vector channels
similarly , if the mass is decreased ( compare figs .
13 and 14 ) the difference between pseudoscalar and vector correlators after cooling is enhanced .
although our lattices are too small to do very accurate mass determinations we were able to get reasonable fits to the pseudoscalar correlator on the @xmath39 lattice in a restricted time window with the function @xmath55,\ ] ] where @xmath56 is the size of the lattice . the meson decay constant @xmath57 can be calculated @xcite from the coefficient @xmath58 according to @xmath59 .
the results for the mass and decay constant are given in table 1 .
the pattern is consistent with the findings of chu _ et al .
the masses ( in lattice units ) are reduced to about 2/3 of the uncooled values and do not change very much in the range of intermediate cooling ( 20 to 50 cooling steps ) .
the meson decay constant , which reflects the short - distance behavior of the wave function , is reduced to about 1/2 of the uncooled value .
how is one to interpret the behavior of the meson correlators under cooling ? one way is that the correlators reflect the true dynamics of the cooled vacuum . at sufficiently large times and
sufficiently small masses the correlation functions are not affected by cooling very much so hadron properties ( and the uncooled lattice scale ) survive essentially intact .
this is the approach of chu _ et al .
an obvious question with this interpretation is how to reconcile it with the finding of direct calculation that the spin - dependent potentials and chiral symmetry breaking essentially disappear ( in lattice units ) after only a small amount of cooling .
another interpretation is that the effect identified by teper @xcite in the string tension is also present in the meson correlators .
namely , correlations functions do not relax uniformly under cooling . at sufficiently large time separations
it is inevitable that the behavior of the original uncooled configurations persists and this does not reflect the dynamics of the cooled vacuum .
in this work the effect of cooling on a number of observables was calculated in su(2 ) lattice gauge theory .
these include the central and spin - dependent potentials , the chiral symmetry breaking order parameter and meson correlators .
even after 100 cooling steps a remnant of the confining static potential is seen at large distance .
however interactions on the distance scale of a few lattice spacings are essentially eliminated .
this includes the spin - dependent interactions induced by chromo - magnetic field correlations .
the quantity @xmath0 was found to approach its free - field value very quickly with cooling .
this may indicate that on the lattice , with staggered fermions , chiral symmetry breaking is driven more by local fluctuations in the topological charge density rather than by global topological properties . as cooling smooths out local fluctuations , chiral symmetry breaking decreases rapidly , even though instantons remain .
this is in line with the study of hands and teper @xcite who suggest that the instanton - induced zero modes for staggered fermions do not become delocalized as would be required for chiral symmetry breaking .
meson correlators up to time separations of several lattice spacings relax quickly to free - field values reflecting the behavior of the potentials . at larger times , differences from free - field behavior persist .
a fit to the pseudoscalar correlation function yields `` mass '' values about @xmath1 of the uncooled masses . to this extent
our results confirm the calculation of chu _ et al .
_ @xcite .
on the other hand , we can not conclude that our results provide evidence for the dominant role of instantons .
there are properties of qcd that are changed by cooling .
the lack of a direct signature for spin - dependent forces and chiral symmetry breaking has to be reconciled with the large - time behavior of the hadron correlators . until this can be done the interpretation of the hadron properties after cooling remains imprecise .
r. scalettar , d. scalapino and r. sugar , phys .
b * 34 * , 7911 ( 1986 ) ; s. gottlieb , w. liu , d. toussaint , r.l .
renken and r.l .
sugar , phys .
d * 35 * , 3972 ( 1987 ) ; k. bitar , a.d.kennedy , r. horsley , s. meyer and p. rossi , nucl .
b313 * , 348 ( 1989 ) . | the effect of cooling on a number of observables is calculated in su(2 ) lattice gauge theory .
the static quark - antiquark potential and spin - dependent interactions are studied , and the topological charge is monitored .
the chiral symmetry breaking order parameter @xmath0 and meson correlators are calculated using staggered fermions .
interactions on the distance scale of a few lattice spacings are found to be essentially eliminated by cooling , including the spin - dependent potentials .
@xmath0 and meson correlators up to time separations of several lattice spacings relax very quickly to their free - field values . at larger times
, there is evidence of a difference between the pseudoscalar and vector channels .
a fit to the pseudoscalar correlation function yields `` mass '' values about @xmath1 ( in lattice units ) of the uncooled masses .
these results raise the question of how to reconcile the large - time behavior of the hadron correlators with the fact that the spin - dependent potentials and @xmath0 essentially disappear ( in lattice units ) after only a small amount of cooling . |
to understand the deep structure of loop quantum gravity ( lqg ) @xcite , in particular the holographic principle and quantum black holes , it is necessary to analyze in detail the entropy counting .
most of the work on black hole entropy in lqg focuses on boundary state counting , especially in the isolated horizon framework @xcite . in the present work we propose to extend these considerations to the bulk entropy .
we use the framework outlined in @xcite . we consider a given spin network state for the 3d geometry of the ( canonical )
hypersurface and focus on an arbitrary bounded region . the previous work @xcite analyzed the boundary entropy of such a region assuming a totally mixed state , i.e having no knowledge about its interior .
we naturally recovered the area - entropy law . in the present work ,
we extend this calculation to the bulk entropy .
more precisely , we retain the information about the graph ( underlying the quantum geometry state ) inside the considered region and , under given boundary conditions , count all the distinct spin network states on that graph
. then the entropy defined as the logarithm of the number of states depends on the topology of the graph through its number of loops . in the trivial topology case with no loop
, we recover the previous result obtained by counting boundary states @xcite .
as soon as the topology becomes non - trivial the entropy diverges .
nevertheless , we identify a symmetry responsible for this divergence and get a finite entropy after suitable gauge fixing .
the entropy increases with the number of loops . for a complicated enough graph topology
we find that the entropy can grow arbitrarily large compared to the boundary area .
however , we also identify a regime for which the entropy still scales as the area .
we call it the holographic regime " .
there it turns out to be possible to arbitrarily adjust the proportionality factor between the entropy and the area ( without changing the immirzi parameter ) .
if quantum gravity is to be a holographic theory , then the lqg dynamics should be such that projecting on physical states selects this regime .
finally , the gauge fixing that we use seems to be related to a gauge fixing of the hamiltonian constraint , but more investigations are required to understand that relationship .
let us start with an arbitrary spin network state on the canonical hypersurface @xmath0 . consider a connected bounded region @xmath1 of its graph that includes a finite set of vertices and the edges that connect them .
the boundary @xmath2 is the set of edges which have only one end vertex laying in @xmath3 .
we can picture @xmath3 as a 3-ball and @xmath2 as its boundary 2-sphere punctured by the boundary edges .
, width=188 ] let us call @xmath4 the graph inside @xmath3 and assume that the boundary @xmath2 is made of @xmath5 edges .
the spin network state carries spin labels ( @xmath6 representations ) attached to each edge of the graph .
in particular , the boundary data consists of the spin labels @xmath7 of the @xmath5 boundary edges puncturing @xmath2 .
we assume these labels fixed once and for all as defining the boundary .
we further take them all equal to the lowest spin @xmath8 . as it was shown in @xcite ,
this hypothesis simplifies the calculations and it is straightforward to generalize the calculations to any generic configuration .
for such a choice of spins the number of boundary edges @xmath5 is necessarily even , and we therefore assume in the following that the boundary is made of @xmath9 edges .
in @xcite we motivated defining boundary states as intertwiners between the @xmath9 boundary representations or equivalently as singlet states in the tensor product of the @xmath9 spin @xmath10 .
counting the dimension @xmath11 of the intertwiner space gives the ( boundary ) entropy @xmath12 in term of the binomial coefficient @xmath13 : s _ n_= 2n2 -32n + assuming that every @xmath14 puncture contributes a microscopic area @xmath15 in planck units , we find that the entropy satisfies the usual proportionality law at the leading order , and has a @xmath16 logarithmic correction : s_~ -32a+ , where the total boundary area is @xmath17 . in the standard lqg framework , the microscopic area is given in terms of the immirzi parameter @xmath18 . then the semi - classical relation @xmath19 can be used to fix the value of the immirzi ambiguity @xmath20 .
the isolated horizon framework presents a different though similar calculation for the entropy .
it leads to a different ratio @xmath21 and thus a different value of @xmath20 , but also to a different value of the logarithmic correction - usually @xmath22 instead of the present @xmath23 .
this boundary entropy calculation can be interpreted as computing the number of spin network states assuming that the graph @xmath4 inside the region @xmath3 is reduced to a single vertex .
assuming that the outside observer have no access to any information about the graph inside @xmath3 , we have indeed coarse - grained this graph to a single point : this can be dubbed the black point " model . in the present work
we investigate the effects of keeping a non - trivial graph @xmath4 on the entropy calculation . as a spin - network wave functional is a function of the ( gravitational ) holonomies , it is possible to have an impression that degrees of freedom are attached to the edges of @xmath4 .
however , due to the gauge invariance , degrees of freedom are truly carried by the loops of the graph .
consider @xmath4 with @xmath24 vertices , @xmath25 internal edges and @xmath26 external legs .
then the number of independent loops is @xmath27 .
when @xmath4 is a tree , @xmath28 , the entropy is exactly the same as for the trivial graph with a single vertex .
as soon as the graph possesses a non - trivial topology , @xmath29 , we obtain an infinite number of states and thus of degrees of freedom .
indeed , loops of a spin network can carry arbitrary spins @xmath30 independently of each other . to overcome this obstacle
, we note that this divergence is associated to a further gauge invariance .
this gauge invariance is generated by the action of the holonomy operators on the loops of the spin network state . to count physical degrees of freedom and to get a meaningful finite entropy , we gauge fix this action @xcite .
we use the simplest gauge fixing and fix the spin associated to each loop of @xmath4 .
it actually requires fixing the spin of only one link of each loop .
having done this , we will see in the following section that we indeed obtain a finite dimensional hilbert space .
the resulting entropy does not depend of the choice of the spins used in the gauge fixing and only depends on the size / area of the boundary @xmath26 and on the complexity of the graph @xmath4 described by the number of loops @xmath31 .
this gauge fixing of the holonomy operators on the graph loops can be related to the gauge fixing of the hamiltonian constraint . in the context of the topological bf theory ,
this would actually exactly coincide with the action of the hamiltonian constraint . in the case of lqg , it is more subtle and we should investigate further the validity and physical interpretation of our gauge fixing . while we postpone this for future work , we insist that in our framework this gauge fixing is natural .
first , we gauge fix the holonomies on loops inside the considered region @xmath3 .
this a priori does not affect the dynamics of the exterior of @xmath3 .
on the other hand , if we did not gauge fix them we would be over - counting the number of states ( seen by an external observer ) .
what would be non - trivial is to gauge fix the action of holonomies on loops crossing the boundary / horizon @xmath2 .
second , we do check that the number of degrees of freedom does not depend on the gauge fixing and that the action of the holonomies creates isomorphic copies of the same physical " hilbert space . in short
, we are disregarding the internal excitations that do not couple to the exterior of the region and that would produce an overcounting in the entropy ascribed to @xmath3 by an external observer . in the following sections we study in detail the one - loop case and
then show that the calculations can be straightforwardly generalized to an arbitrary graph topology .
finally we discuss the different regimes of entropy corresponding to the different scaling of the number of loops @xmath31 with the size of the boundary @xmath5 .
let us start by reviewing the black - point case , @xmath28 .
the hilbert space of spin networks of a tree is isomorphic to the space of intertwiners between the boundary representations @xcite .
its dimension is given by : _ 0 = _ ( 2 ) dg _ 12(g)^2n= , where @xmath32 is the normalized haar measure on the @xmath6 lie group and @xmath33 is the character in the fundamental spin-@xmath34 representation .
this formula can be interpreted in terms of a random walk with a mirror in origin .
the binomial coefficients @xmath13 give the number of returns to the origin for the usual random walk , while the mirror introduces the factor @xmath35 .
consider now the one - loop diagram on the left of fig .
[ oneloop ] .
it is a loop to which @xmath9 legs are attached , so all the vertices are 3-valent gauge invariance at every vertex ( see e.g. @xcite . ] .
pick an arbitrary link @xmath36 and fix the spin which it carries , @xmath37 .
this obviously fixes the action of the holonomy operator on the loop : acting with a holonomy carrying a spin @xmath38 on this state would change the spin-@xmath39 representation into the tensor product representation @xmath40 .
, width=264 ] assume that @xmath41 . moving from the initial link @xmath36 to the next link @xmath42 , we see that @xmath42 can carry either the spin @xmath43 or @xmath44 since the external leg between @xmath36 and @xmath42 injects a spin @xmath34 into the diagram .
going on one finally arrives back to the initial link @xmath36 and the initial representation @xmath39 .
therefore , the number of possible spin labeling @xmath45 is exactly the number of returns to the origin of a random walk after @xmath9 iterations : _ 1=2nn .
we get a finite result which is different from the tree case .
moreover , it does not depend on the chosen spin @xmath39 .
this hilbert space leads to the following asymptotic behavior of the entropy : s_1 2n2 - 12n+ it does not affect the leading order proportional to the area but only the logarithmic correction . there is nevertheless a subtlety : the assumption that the spin @xmath39 is large enough compared to the boundary size @xmath5 . indeed
if @xmath39 is smaller than @xmath46 , there exists the possibility that the random walk along the loop will ascribe a spin 0 to a link .
in such a case , we can not move down to a @xmath47 spin but can only go back up to @xmath48 .
then we get a smaller hilbert space .
we discard this case because we do not consider it as a true one - loop graph anymore .
indeed , if a link is labeled with a spin @xmath49 , it is just as if that link did not exist since the corresponding spin network wave function would not depend on the holonomy on that link .
then if we remove that link from the graph , we end up with a tree again .
therefore , we interpret this situation when @xmath50 as describing a superposition of a 0-loop and 1-loop graphs .
however , we can still compute exactly the dimension of the hilbert space . for this purpose , it is more convenient to use the another one - loop diagram that is shown on the right of fig .
[ oneloop ] .
we now look at the one - loop diagram with two vertices : all the boundary links merge into a single link to which the loop is then attached .
assume that the loop still carries a fixed spin @xmath39 and we call @xmath51 the spin carried by the intermediate link .
on one side , we have an intertwiner between the @xmath9 spin @xmath34 and the spin @xmath51 .
as long as @xmath52 , the dimension of this intertwiner space is @xcite : d^(n)_k=2nn+k-2nn+k+1=. on the other side , we have a unique 3-valent intertwiner between two representations @xmath39 and the same representation @xmath51 .
the corresponding intertwiner space is of dimension one as long as @xmath53 .
the total number of spin network states amounts to summing over all possible @xmath51 s from @xmath49 to the maximal allowed spin @xmath54 : n_1^(j)= _ k=0^k_maxd^(n)_k = 2nn-2nn+ k_max+1 .
when @xmath39 is larger than @xmath46 , we recover the previous result with @xmath55 . at the other end of the spectrum ,
when @xmath39 vanishes , we recover the 0-loop case with @xmath56 .
finally , we see that the values of the spin @xmath39 carried by the loop between @xmath49 and @xmath46 interpolate between the tree case and the 1-loop case .
this allows us to interpret this intermediate situation as a superposition of the 0-loop and 1-loop cases , and moreover to identify the true 1-loop case as the @xmath41 regime when the entropy does not depend anymore on the specific value of @xmath39 . to conclude this section
, we showed that considering a one - loop graph over a tree increases the entropy .
after carefully gauge fixing , we obtain a finite entropy which differs from the 0-loop case only by the logarithm correction , @xmath47 instead of @xmath16 . in the following sections
, we generalize this analysis to an arbitrary number of loops
. we will show that if the number of loops is fixed it will only affect the logarithm correction while if we allow the number of loops to scale with @xmath5 it may well also affect the leading order .
we now move to the entropy counting in the two - loop case .
we work with the graph shown on fig.[looptopetal ] and we fix the spins along the two loops to the same value @xmath39 for the sake of simplicity .
we further assume as previously that @xmath41 so that we consider a pure two - loop graph and not a superposition with a 0-loop or 1-loop configuration .
the dimension of the hilbert space of spin networks on @xmath4 is given by the following integral : @xmath57 , width=264 ] for computational purpose , it is convenient to express as a sum over the intermediate spin label @xmath51
. it is straightforward to obtain n_2=_k=0^n d_k^(n ) .
the two terms can be computed exactly : _
k=0^n ( 2k+1)d_k^(n)=2 ^ 2n,_k=0^n k(k+1)d_k^(n)=n2n n ~2 ^ 2n .
as we see we have a residual @xmath39-dependent term , which actually diverges as @xmath39 grows to infinity .
neglecting is indeed be neglected if @xmath39 goes to infinity faster than @xmath58 .
this condition is automatically satisfied since we assume that @xmath41 . ]
the extra - term @xmath59 , the @xmath39-dependence can be factorized .
therefore , we interpret this as a symmetry that we havent gauge fixed yet .
indeed we have gauge fixed the action of the holonomy operator along the loops @xmath60 and @xmath61 by requiring @xmath62 , but the action of the holonomy along the loop @xmath63 is not yet fixed .
this leaves a freedom in the intertwiner which leads to that @xmath64 factor , resulting in a over - counting in the hilbert space dimension . ,
width=128 ] this can easily be seen considering the two - loop diagram as shown in fig .
[ twoloops ] .
@xmath38 can still vary from 0 to @xmath64 .
this is the freedom that requires gauge - fixing .
it is possible to compute the intertwiner dimension for different values of @xmath38 .
it is possible to compute the intertwiner dimension for different values of @xmath38 .
for instance , we get : @xmath65 both these examples , have the same behavior up a factor with a leading order given by @xmath66 .
they give the same entropy in the asymptotic limit @xmath67 ( same leading order and logarithmic correction ) : s_2~2 ^ 2n ~2n 2 + with a vanishing logarithmic correction ( they are of course sublog corrections ) .
we see that adding one loop from the one - loop case corresponds to implementing a @xmath68 factor in the logarithmic correction without affecting the leading order proportional to the area .
we generalize this statement to arbitrary number of loops in the following section .
for a generic graph with @xmath31 loops , following the one - loop and two - loop cases , it is reasonable to expect the gauge - fixed dimension at leading order to scale as n_l~_k=0^n k^l-1d_k^(n ) . [ guess ] assuming that @xmath31 is fixed , the asymptotics is in the large @xmath5 limit only changes the logarithmic correction of the entropy : s_ln_l~2n2 + ( l 2 -1)n + [ asymptguess ] using the same techniques as in @xcite , it is straightforward to approximate the sum over @xmath51 by an integral and then compute the asymptotics of @xmath69 by a saddle point approximation
. then only the logarithmic correction changes by a factor @xmath68 with each loop because the distribution @xmath70 looks like a gaussian peaked at @xmath71 with a width scaling as @xmath58 .
more precisely , denoting @xmath72 $ ] , we can approximate the dimensions @xmath70 for large @xmath5 using the stirling formula : d_k^(n)~ 2 e^-n(x ) , with the exponent @xmath73 given by : ( x)= ( 1+x)(1+x)+(1-x)(1-x ) .
therefore , the full hilbert space dimension can be approximated by an integral : @xmath74 the exponent @xmath73 is always positive for @xmath75 $ ] .
it has a unique fixed point at @xmath76 . for large @xmath5
, we can use the gaussian approximation , @xmath77 .
then we need to evaluate @xmath78 , which scales as @xmath79 .
finally , this leads to the asymptotics for the entropy given above in ( [ asymptguess ] ) .
so far we have kept the number of loops @xmath31 fixed as the boundary area @xmath5 was taken to infinity .
this number does not affect the leading order of the entropy , but only changes the pre - factor of the logarithmic correction .
however , we see that if we allow the graph complexity to grow with the boundary size , we are able to change the leading order behavior of the entropy and change the proportionality factor between the entropy @xmath80 and the area @xmath5 . below we make this argument more precise and compute the dimension and entropy exactly for a specific choice of gauge - fixing .
we find a missing factor @xmath81 in the guessed dimension ( [ guess ] ) .
this changes the asymptotical behavior when @xmath31 is allowed to be arbitrarily large .
then we keep the asymptotics described above for fixed @xmath31 while allowing the loop number @xmath31 to scale as @xmath5 changes the leading order factor between the @xmath82 and @xmath5 .
let us now consider the @xmath31-loop graph with three vertices @xmath83 as shown on fig [ manyloops ] .
the @xmath9 boundary links combine into intermediate spin @xmath51 at the vertex @xmath84 . on the other side ,
the @xmath31 loops combine at the vertex @xmath85 to that same intermediate link .
finally the intertwiner at the @xmath86 node closes the graph and describes how the @xmath31 loops are coupled to each other .
, width=151 ] our gauge fixing " , or more precisely , the choice of a sector of the spin network hilbert space , is achieved by fixing the spin labels on the @xmath87 edges defining the @xmath31 loops .
we label them sequentially by representations @xmath88 . as long as @xmath39 is assumed larger than @xmath46 as before
, this leads to an entropy which does not depend on the gauge - fixing representation @xmath39 but only on the number of loops @xmath31 and the number of boundary edges @xmath5 .
we need to underline that we are not only fixing the action of the holonomy operator along the @xmath31 loops .
indeed , our specific choice of representation labels is not only convenient for computational purposes but also it fully fixes the unique intertwiner at the node @xmath86 . this amounts to fixing @xmath89 representation labels within the node @xmath86 .
this means that we are truly fixing @xmath90 quantum numbers , and not only @xmath31 as we first expected .
we are in fact counting the number of possible intertwiners at the vertices @xmath84 and @xmath85 . at a nave level , the entropy that we compute counts only the different ways that the internal loops couple to the external links while we disregard the number of ways that the loops couple to each other by fixing the intertwiner at the internal node @xmath86 . at a mathematical level , if one fixes less representation labels than we do ( less than @xmath91 ) , one gets an entropy which depends on @xmath39 and diverges as @xmath39 grows large .
this is a symptom of insufficient gauge - fixing .
nevertheless , there remains the open issue of understanding the physical interpretation and relevance of such a gauge - fixing .
in particular , what is the precise symmetry that we are gauge - fixing ?
we leave this question for future investigation . however , even without such an interpretation in term of symmetry and gauge - fixing , we still have identified sectors of the spin network hilbert space with different asymptotic behaviors of the entropy of the considered region which only depend on the number of loops of the graph ( supporting the spin network states ) inside that region . to compute the entropy
, we need to find the number of intertwiners at the vertices @xmath84 and @xmath85 . at the vertex @xmath84
we have the same degeneracy @xmath70 as previously .
as for the vertex @xmath85 , having @xmath31 loops leads to the following degeneracies for the spin-@xmath51 subspaces as long as @xmath92 for @xmath29 : ^(l)_k = k+l-1k . this formula is straightforward to prove by induction using the identity @xmath93 . as a result , the dimension of the spin network space inside @xmath94 is n_l=_k=0^n d^(n)_k ^(l)_k,[ndef ] where we have assumed @xmath95 . for @xmath96
this can be slightly simplified : n_l=_k=0^n d^(n)_k k+l-1k = _
k=0^n 2nn+kk+l-2k,[nicesum ] for a small number of loops , we easily get the exact results for @xmath69 , using properties of the binomial coefficients @xcite . for @xmath97
, we recover the previous result @xmath98 , with an asymptotic expression for the entropy @xmath99 . for @xmath100
one obtains n_2=_k=0^n2nn+k=2
^ 2n-1 + , with the entropy @xmath101 having no logarithmic correction for large @xmath5 ( of course , it still contains sub - logarithmic corrections ) . for @xmath102
, we can also compute : n_3=_k=0^n2nn+k(k+1)=+2 ^ 2n-1 , with a logarithmic correction being @xmath68 . in the analysis of the entropy asymptotics
, we distinguish between three cases .
first , if the number of loops @xmath31 is held fixed ( or more generally @xmath31 stays negligible compared to @xmath5 ) , we recover the results given above : the leading order @xmath103 does not change , while the non - trivial topology of the graph affects the logarithmic correction which becomes @xmath104 . in the second case
@xmath31 is scaling as @xmath5 : the entropy depends on the limit of the ratio @xmath105 as @xmath5 goes to infinity . in the third case @xmath31 grows very large compared to @xmath5 : if the ratio @xmath106 is unbounded , the leading order of the entropy drastically changes and will not scale proportionally to @xmath5 anymore .
the asymptotic expression for all the above cases can be derived from a saddle point approximation of the sum @xmath107 as shown in details in appendix .
it also allows to obtain the sub - leading terms with an excellent precision . when @xmath108 we get a compact expression , s_l = n_l , n_l~,[lsmall ] where @xmath109 .
this still gives a good estimate of the sub - leading terms : for example , when @xmath110 and @xmath111 , the difference between the exact and asymptotic values is @xmath112 , while @xmath113 .
when @xmath114 with a fixed ratio @xmath115 , we can still apply the same saddle point technique . using stirling formula , we approximate the dimension @xmath69 by the integral expression , n_l~ _ 0 ^ 1 dxe^-n(x ) , with the new @xmath115-dependent exponent , ( x)= ( x)+xx -(x+)(x+ ) .
as shown in details in appendix , the gaussian approximation controls the asymptotic behavior of the entropy : s_l = n~ n -12n + the important new result is that the leading order factor @xmath116 depends on @xmath115 and is not fixed to the standard factor @xmath117 anymore .
we show on fig .
[ l0l1l2 ] a plot with maple numerics for the entropy @xmath118 for @xmath28 ( as a reference ) and @xmath119 .
while the leading behavior is always linear , the slop @xmath120 clearly depends on the value of the ratio @xmath121 .
the final case is when the complexity of the graph @xmath31 grows much faster than the boundary size @xmath5 . in this case
, the entropy grows faster than linearly compared to the boundary area : s_ln~
n(l - n ) + n -12n+
after a suitable gauge fixing , we have managed to compute the finite bulk entropy counting the number of spin network states supported by a fixed graph @xmath4 inside the considered bounded region @xmath94 with fixed boundary conditions .
the entropy @xmath80 only depends on the number of links @xmath5 puncturing the boundary @xmath122 and the number of loops @xmath31 of @xmath4 which quantifies the complexity of the graph .
computing the asymptotics of @xmath82 , we distinguish three regimes . *
* the logarithmic regime :* + the number of loops @xmath31 is held fixed while @xmath5 is take to infinity .
the leading order of @xmath82 is linear in @xmath5 and is exactly the same as for the boundary entropy @xcite .
the complexity of the graph only affects the factor in front of the logarithmic correction @xmath123 , which increases with @xmath31 .
therefore , if a ( quantum ) black hole is belongs to this regime , fixing the smallest spin @xmath124 fixes the immirzi parameter . *
* the holographic regime :* + the number of loops @xmath31 scales proportionally to the horizon area @xmath5 .
the entropy still grows linearly in @xmath5 in the leading order but the proportionality factor changes and depends on the ratio @xmath121 .
this is a sector of the spin network space where the entropy still scales with the boundary area and not faster .
if the full lqg theory is to be ( strongly ) holographic , then the dynamics should restrict the physical states to this regime : the graph complexity can not grow faster than the boundary size . on the other hand ,
even if the lqg theory is not restricted to this regime , a quantum black hole state should lay in this regime if we want to respect the semi - classical area - entropy law .
then since the factor @xmath125 depends on the ratio @xmath115 , we can adjust this new parameter @xmath115 so to get @xmath126 without changing neither the immirzi parameter nor the area quantum @xmath127 . in this scenario ,
the black hole entropy calculation does not fix the immirzi parameter . from another perspective
, we can say that the factor @xmath115 renormalizes the immirzi parameter @xmath20 .
finally we need the dynamics to select the physical sector of the spin network space and provide us with the
right " value of @xmath115 .
of course , this scenario only holds if we use the bulk entropy and do not restrict ourselves to work with the boundary entropy . * * the non - linear regime :* + if the number of loops grows much faster than @xmath5 , i.e @xmath128 , then the entropy is free to grow non - linearly with @xmath5 . since the leading order is @xmath129 , we can , e. g. , get an entropy scaling with the volume " @xmath130 for an exponential growth of the type @xmath131 . of course
, one should keep in mind that the volume of @xmath94 actually depends on the bulk state and does not always scale as @xmath130 .
it could well be larger ( or smaller ) if the space is tightly curved .
we explored the notion of bulk entropy " in the framework of loop quantum gravity , in order to generalize the calculations of boundary entropy of @xcite .
our aim was to compute the number of ( spin network ) states describing the quantum geometry of a bounded region of space , living on a fixed graph and with fixed boundary conditions .
this bulk entropy is obviously related to the complexity of the graph and was found to be related to the number of loops @xmath31 of that graph .
the straightforward calculation then gives at first an infinite entropy .
nevertheless , after fixing a certain number of labels of the spin network states , we were able to extract meaningful finite results .
this lead to the identification of different sectors of the spin network hilbert space depending on @xmath31 with different asymptotical behavior of the entropy .
we can a priori have an entropy growing as large as we want if we choose a number of loops @xmath31 large enough .
nevertheless , two sectors are particularly interesting .
if the graph complexity is fixed while the boundary area grows , then the bulk entropy @xmath80 has the same leading order than the boundary entropy but the logarithmic correction to the entropy changes and increases with @xmath31 . on the other hand , if @xmath31 grows linearly with the boundary area @xmath84 , then the bulk entropy also grows linearly with the area but the ratio @xmath21 depends explicitly on the ratio @xmath132 .
we call this sector the holographic regime "
. if we want black holes to satisfy the area - entropy law , the quantum black hole state must necessarily be in the holographic regime .
however , this regime is generic enough to allow for a fine - tuning of the ratio @xmath21 to @xmath133 without changing the immirzi parameter ( i.e the value of the minimal quantum of area ) .
indeed , we can always change the area - complexity ratio @xmath132 to adjust the value of @xmath21 .
our point of view is that it is the lqg dynamic that is supposed to select the physical regime and value of @xmath132 .
there are two important issues to address in this scenario .
first , should the considered entropy for the space region be its boundary entropy or its bulk entropy ?
we see a priori no reason why the degrees of freedom inside the region would not couple to the space outside
. then the bulk degrees of freedom would affect the black hole evaporation .
however , the case of a black hole might ( should ? ) be drastically different from a generic region of space . this is a question that needs to be addressed by the lqg dynamics . the second issue is directly related to our approach .
our entropy calculation relies on a partial fixing of the spin network labels .
we discussed that it could be interpreted as a gauge fixing of the hamiltonian constraint , but this needs to be worked out in details .
thus , the limits of validity of the present work depend on understanding the legitimacy of this gauge fixing .
however , even if it does not turn out to be interpreted as the gauge fixing of a certain symmetry , we have nevertheless identified the different sectors of the spin network kinematical space with different bulk entropy behavior , but we still need to provide these different sectors with a proper physical interpretation .
the asymptotic expressions for @xmath134 and the different regimes of @xmath135 are based on the use of stirling formula , replacement of the sum by the ( euler - maclaurent ) integral and the saddle point ( gaussian ) approximation .
it is convenient for the analysis to introduce a shifted loop number @xmath136 and consider the ratio @xmath137 . the first factor in the sum of eq .
( [ nicesum ] ) takes the form ~f(k
/ n)e^-n(k / n ) , where f(x)= , ( x)= ( 1+x)(1+x)+(1-x)(1-x ) .
the second factor becomes ~=g(k / n , l / n)e^-n(k / n , l / n ) , where g(x,)= , ( x,)=xx -(x+)(x+ ) .
the sum of eq .
( [ nicesum ] ) is replaced by the integral n_l~ _ 0 ^ 1 dx f(x)g(x,)e^-n(x , ) , with @xmath138 .
the saddle point approximation requires the knowledge of the derivatives of the exponent : _ x=,_x^2=. the unique fixed point , @xmath139 , in the interval @xmath140 $ ] is x_0(- ) , and it is easy to check that @xmath141 is always positive on @xmath140 $ ] . in particular , for @xmath142
the minimum of @xmath143 is inside the interval and we use the gaussian approximation @xmath144 that gives the following asymptotics for the entropy : s(l , n)=n_l , n_l~f(x_0)g(x_0 , ) .
[ gauss ] as a matter of fact , this approximation is in an excellent agreement with the exact results : e. g. , @xmath145 and the error is @xmath146 , increasing only to @xmath147 . at the other extreme , @xmath148 with @xmath149 .
when @xmath115 runs from 0 to @xmath150 , the fixed point @xmath151 varies from 0 to 1 . in the special case @xmath152
when @xmath153 , @xmath154 and we get simple asymptotics : s(l = n)~ 3n2 -12n-12 . in this case , we can actually give a faster proof of the asymptotics : @xmath155 in the limit regime @xmath156 , we have @xmath157 which improves the estimate of eq .
( [ lsmall ] ) to s(ln ) & = & 2n2 + n - l + ( 1 - 2 ) - l -(4 ) + & & - + l+ - - + ( ) . finally , in the case that @xmath158 , we have @xmath159 and eq .
( [ gauss ] ) yields a simple expression for the entropy , s(ln)~nl-(n+)n+n-2+(6n-1)/12l+(n^2/l^2 ) .
99 c. rovelli , _ loop quantum gravity _ , ( cambridge university press , cambridge , 2004 ) ; + t. thiemann , _ lectures on loop quantum gravity _ , lect
. notes phys .
* 631 * 41 - 135 ( 2003 ) , arxiv : gr - qc/0210094 ; + a. ashtekar and j. lewandowski , _ background independent quantum gravity : a status report _ , class .
* 21 * ( 2004 ) r53 , arxiv : gr - qc/0404018 . | abstract in the framework of loop quantum gravity ( lqg ) , having quantum black holes in mind , we generalize the previous boundary state counting ( gr - qc/0508085 ) to a full bulk state counting . after a suitable gauge fixing
we are able to compute the bulk entropy of a bounded region ( the black hole " ) with fixed boundary .
this allows us to study the relationship between the entropy and the boundary area in details and we identify the holographic regime of lqg where the leading order of the entropy scales with the area .
we show that in this regime we can fine tune the factor between entropy and area without changing the immirzi parameter . |
in recent years a growing interest has been focused on crystalline systems lacking inversion symmetry ( see for example ref . ) . a direct consequence of non - centrosymmetry is the occurrence of a spin - orbit dzyaloshinsky - moriya ( dm ) interaction @xcite .
the dm interaction is thought to play an important role in multiferroic systems and may be at play in the occurrence of non - trivial magnetic structures . on the other hand , non - centrosymmetric systems presenting superconductivity have attracted special attention as the absence of parity symmetry considerably restricts the possible superconducting states ( i.e. a pure triplet pairing is in principle not allowed without inversion symmetry ) @xcite .
mnsi is probably the non - centrosymmetric system which concentrates most of the research focus .
it was the first material found to exhibit a homochiral spin spiral structure below @xmath3 k @xcite .
this spin spiral adopts a very long wavelength of about 18 nm .
in addition to an external pressure destroying the magnetic state ( @xmath4 gpa @xcite ) , a modest magnetic field of 6 kg is found to overcome the dm interaction leading to a spin - aligned state @xcite .
moreover a skyrmion phase was recently identified near @xmath2 in the magnetic phase diagram @xcite .
contrary to what observed in a number of strongly correlated electron systems ( see for example ref . ) , mnsi does not display superconductivity at @xmath5 .
it was invoked @xcite that the absence of a superconducting state for mnsi under pressure has to be traced back to the absence of parity operator for the wavefunctions . in view of all these properties
, mnsi has been investigated several times by @xmath0sr in zero or externally applied pressures ( see for example refs . ) .
two main controversies occurred from these studies .
the first one regards the behavior of the magnetic state near the critical pressure @xmath5 .
first measurements performed by uemura _ et al . _
@xcite seemed to show a phase separation between magnetic and paramagnetic regions whereas measurements performed by andreica _ et al . _
@xcite , extending at lower temperatures , showed a complete absence of phase separation .
the second controversy regards the muon stopping site(s ) and the occurrence or not of magnetic polaron .
the large majority of the studies associates the two spontaneous frequencies seen below @xmath2 to two magnetically inequivalent sites .
the same is deduced from transverse - field ( tf ) measurements performed at low fields @xcite , even though the conclusions obtained from this latter study are blurred by the assumption of a vanishing dipolar tensor . on the other side , storchak _ et al . _
@xcite relate the observation of two frequencies in tf data to a coupled @xmath6 system ( i.e. mu state ) . in this scenario ,
the spin of the mu electron does not exhibit rapid fluctuations through spin exchange with the itinerant electrons , but is kept fixed " by the local ferromagnetic ordering mediated by itself ( the electron and the four mn neighbors are thought to be coupled and to behave as a single entity with large spin : i.e. a spin polaron " ) . in this scenario
, the two frequencies correspond to two muon spin - flip transitions between states with fixed electron spin orientation .
the picture of such a localized electron state appears controversial given the metallic nature of mnsi .
the aim of the present paper is to lift the controversy regarding the occurrence or not of magnetic polaron and provide information on the muon stopping site .
we retain this an essential point to solve , as this will show whether @xmath0sr can be utilized to specifically extract quantitative information on this heavily studied strongly correlated system . in this study
, we will show that the occurrence of different frequencies in the tf-@xmath0sr spectra can be traced back to the different responses of crystallographically equivalent muon - stopping sites which become magnetically inequivalent in the presence of an external magnetic field and that one does not need to invoke a magnetic polaron to explain the tf-@xmath0sr data .
we will also show that the occurrence of two spontaneous muon frequencies in the zero - field @xmath0sr data recorded below @xmath2 can be explained within this frame . finally , we also present density - functional theory calculations which perfectly support the experimental evidences .
we used the same mnsi sample as in the previous @xmath0sr under pressure experiments @xcite .
it consists of a single crystal with the form of a cylinder of approximately 7 mm diameter and 19 mm length ( see fig . [ sample ] ) .
the single crystal was grown by the czochralsky pulling technique from a stoichiometric melt of high - purity elements ( @xmath7% ) using radio - frequency heating and a cold copper crucible .
no deviation from the known crystal structure [ space group @xmath8 , no .
198 ; mn - ion at the position @xmath9 and si - ion at the position @xmath10 ; lattice constant 4.558 ] nor any presence of foreign phase was detected by scanning electron microscope microanalysis , backscattered electron imaging , and x - ray debye scherrer pattern .
sr measurements .
the sample was rotated along the cylinder axis . ]
the study of the angular dependence of the muon knight - shift in mnsi was performed using the gps instrument located at the @xmath11m3 beamline of the hipa complex at the paul scherrer institut ( psi , villigen , switzerland ) .
the measurements were performed with the transverse - field geometry with a magnetic field of 5200 g. the sample was rotated around the cylinder axis and the external field @xmath12 was applied perpendicular to the cylinder .
note that the sample was not grown along a specified crystallographic orientation , i.e. the cylinder axis does not correspond to a principal axis .
the measurements presented here were performed using a dynamical he - flow cryostat ( i.e. sample in he - flow ) at a temperature of 50 k. the temperature stability of the sample was better than 0.1 k for all the measurements .
for all rotation angles @xmath13 , the @xmath0sr signals are best modeled by the presence of four frequencies with different angular dependences .
fits with all parameters free point to similar amplitudes for the four components .
hence , in a second series of fits the amplitudes were forced to be the same .
that is : @xmath14 sr time spectrum plotted in a rotating reference frame ( rrf ) with frequency 65 mhz .
the grey line represents the fit to the data ( open points ) using eq .
( [ asymtheor ] ) and the four individual signals are also displayed ( color code is the same as on fig .
[ raw_data ] ) .
note that the actual fits were performed without rrf and without time binning ( i.e. original bin of 0.9766 ns ) .
notice that the four signals have very similar depolarization rates . ]
the frequencies @xmath15 reflect the local magnetic field values sensed by the muon at the stopping site(s ) , i.e. @xmath16 , where @xmath17 is the gyromagnetic ratio of the muon .
the phase parameter @xmath18 is common to all signals as it is defined by the direction of the muon spin with respect to the detector system at the muon implantation time . during the fits ,
no restrictions were applied concerning the frequencies and the results were directly obtained through the software musrfit @xcite . note that all the fits were performed in the time - space between 0 and 8 @xmath0s . as illustrations to the data , we present in fig . [ raw_data_time ] a @xmath0sr time - spectrum , plotted in a rotating reference frame , and in fig .
[ raw_data_fft ] selected fast fourier transform ( fft ) of spectra recorded at different angles .
the full angular dependence of the frequencies is shown in fig .
[ raw_data ] . and [ raw_data ] .
] sr frequencies .
the lines are guides to the eye .
the green dash - line represents the average frequency @xmath19 ( see text ) . ]
the occurrence of four frequencies in the @xmath0sr signal clearly points to a muon site located at the @xmath20 wyckoff position of the cubic structure . in this view
, the main contribution of the observed angular dependence arises from the dipolar contribution .
as shown in table [ diptensor ] , which provides information on this type of sites and the respective dipolar tensor forms , taking into account the symmetry of the dipolar tensors , the four crystallographically equivalent 4@xmath1 sites in the unit cell are either : i ) magnetically equivalent when @xmath12 is applied along a principal direction ; ii ) grouped in two pairs of magnetically inequivalent sites for @xmath12 applied in a principal plane ; iii ) usually all magnetically inequivalent for @xmath12 applied along an arbitrary direction . [ cols="<,^,^,^,^",options="header " , ] at this point , these results can be used to solve eq .
( [ eqrot ] ) for each signal ( site ) and obtain the euler angles @xmath21 and @xmath22 of the rotation axis , as well as the parameter @xmath23 characterizing the representation of the dipolar tensors @xmath24 in the reference frame of the crystal ( see table [ diptensor ] ) .
as each signal should provide us with the same parameters , we have a rather precise determination of these parameters , i.e. @xmath25 mole / emu , @xmath26 and @xmath27 .
we note that the values of @xmath21 and @xmath22 could be confirmed _ a posteriori _ by performing x - rays laue measurements of the crystal .
but we stress that the values of @xmath21 and @xmath22 extracted by @xmath0sr must be considered as more precise as , for example , they take into account any possible misalignment of the sample mounting compared to the direction of the external magnetic field .
the experimentally determined value of @xmath23 can be directly compared with dipolar sum calculations .
figure [ adip_theor ] , which exhibits the theoretical value of @xmath23 calculated for different sites 4@xmath1 , indicates that this experimental value is compatible with muons sitting either at the site @xmath28 or @xmath29 .
we stress that a small uncertainty on the magnetic susceptibility ( and therefore finally on the obtained value of @xmath23 ) will only weakly affect the determination of the muon - site ( for example , an uncertainty as high as @xmath3010% on @xmath31 leads to a shift of @xmath32 on the muon - site coordinates ) .
the obtained sites correspond to almost symmetrical positions along the crystal diagonal on each side of the plane formed by three adjacent mn - ions ( see fig .
[ structure ] ) .
however , the site @xmath29 appears not probable as located rather close to the si - ion on the crystal diagonal ( @xmath33 ) .
note that such a short @xmath0-ion distance can sometimes be observed when a bounding between the muon and a negative ion occurs , as in the high-@xmath34 cuprates . characterizing the representation
@xmath35 $ ] of the dipolar tensors in the crystal reference frame for the 4@xmath1 sites .
the divergence occurs at the 4@xmath1 position of the mn - ion .
the red dot - line corresponds to the results of the fits as explained in the text . ]
is also indicated ( red ) as well as the other three equivalent sites .
note that six mn - ions , which do not belong to the primary unit cell , are also displayed . ] to demonstrate the perfect agreement between theoretical calculations and experiment , the lines on fig .
[ bdip_corrected ] represent the dipolar calculations for the site @xmath28 with the euler angles @xmath36 and @xmath37 . note that as the four sites are crystallographically equivalent , one expects similar amplitudes of the four signals , as experimentally observed .
we note also that our measurements performed with angles of @xmath38 and @xmath39 correspond respectively to an external field applied almost into a principal plane [ i.e. ( 110 ) ] and almost along the principal axis @xmath40 $ ] ( as the euler angle for the rotation axis is @xmath36 ) . figure [ raw_data_fft ] shows that the signals tend to collapse into two pairs in the first case and all together ( and with a dipolar contribution tending to zero ) in the second case , as expected . for completeness
, we would also like to mention the possibility of a small lattice dilation around the muon .
we notice first that such effects are usually extremely small in metals ( of the order of the percent , if detectable at all ; see for example ref . ) .
a hypothetical weak dilation of the lattice would decrease the calculated absolute values of the @xmath23 parameter , which would push slightly the muon position in the direction of the plane formed by the three adjacent mn - ions ( see fig .
[ structure ] ) .
however , the main conclusion that the muon is stopped at a 4@xmath1 position would not be affected .
finally , we note that a by - product of our analysis is the determination of the contact field @xmath41 .
basically two interactions contribute to the contact field .
the first one is a result of the pauli paramagnetism of the conduction electrons and their fermi contact interaction with the muon @xcite . in systems with localized moments , as mnsi , an additional contribution arises as the spin - polarization of the conduction electrons at the muon site will be increased through the ruderman - kittel - kasuya - yosida interaction due to the local magnetic moments . as the pauli susceptibility of the conduction electrons is much smaller than the magnetic susceptibility from the local @xmath42 moments ( i.e. @xmath43 ) ,
the fermi interaction can be safely neglected in the contact field , which therefore can be written as @xmath44 where @xmath45 is the hyperfine contact coupling tensor . by writing eq .
( [ eq_bloc ] ) , we note that we implicitely assumed that the contact field is independent of the direction of @xmath46 ( i.e. @xmath47 ) as observed in a number of compounds with local moments ( see for example ref . and references therein ) . in any case ,
even if one assumes that the weak angular dependence of @xmath48 is solely arising from a hypothetical angular dependence of @xmath45 , this would correspond to a variation of less than 4% of the hyperfine contact coupling tensor during the rotation . as we _ know _ that some angular dependence has to arise from the demagnetization factor ( see fig .
[ sample ] ) , this number has to be considered as an upper limit . by assuming that the hyperfine contact coupling tensor is isotropic , and taking into account that for our sample geometry the average value of the demagnetization factor is @xmath49 one obtains a value of @xmath50 mole / emu , which can now be included in the computation of the expected values of the spontaneous fields occurring at the muon sites below @xmath2 as described in the next section .
at this point , having a solid knowledge of the muon stopping site , it appears legitimate to thoroughly discuss the number and values of the spontaneous @xmath0-frequencies observed below @xmath2 @xcite in zero - applied field ( zf ) @xmath0sr experiments .
all the @xmath0sr studies reported to date reveal the occurrence of two spontaneous @xmath0-frequencies with values @xmath51 mhz and @xmath52 mhz for @xmath53 k. our first task here will be to discuss why solely two frequencies are observed .
the magnetic structure of mnsi is characterized by spins forming a left - handed incommensurate helix with a propagation vector @xmath54 @xmath55 in the [ 111 ] direction @xcite .
the static mn - moments ( @xmath56 for @xmath57 k ) point in a plane perpendicular to the propagation vector .
the period , which is incommensurate to the lattice constant , is about 18 nm . due to the incommensurability of the magnetic structure ,
one expects a continuous set of local fields at our 4@xmath1 sites , and therefore the field distribution @xmath58 at each site must be considered ( here the index @xmath59 distinguishes the four sub - sites 4@xmath1 ) .
it was shown that such a magnetic structure leads to a field distribution given by @xcite @xmath60 and is characterized by two peaks due to the minimum and maximum cutoff field values .
hence , at a first glance , one would expect to observe up to eight peaks in the fourier spectra of the zf-@xmath0sr data .
however , as the sites @xmath61 , @xmath62 and @xmath63 are located symmetrically around the direction [ 111 ] of the propagation vector ( which represents for them a 3-fold symmetry axis ) , it can be shown that muons stopping at these sites will sense identical field distributions , given by eq .
( [ eq_fd ] ) , i.e. @xmath64 . on the other hand , muons stopping at the site @xmath65 will all sense a unique field value irrespective to the phase of the helix at the muon stopping site .
hence , purely geometrical considerations already reduce the maximum number of peaks in the fourier spectra of the zf-@xmath0sr data down to three . note that this conclusion is not affected if the sample is not monodomain .
hence for the other possible magnetic domains ( which , by taking into account the helicity , are characterized by propagation vectors along [ @xmath66 , [ @xmath67 and [ @xmath68 ) one obtains the same field distributions as for the original domain , albeit the symmetry of the sites with respect to the magnetic structure will be accordingly permuted . ) ] and approximated [ red line : see eq .
( [ eq_fd_approx ] ) ] field distribution expected at the sites @xmath61 , @xmath62 and @xmath63 .
lower panel : muon polarization function deduced from the field distributions shown on the upper panel ( see also text ) . ] to gain more insight , we have computed the expected theoretical cutoff field values of the field distributions . for a given type of muon site , the spontaneous local field in the incommensurate magnetic phase is given by@xcite @xmath69 as before , the index @xmath59 distinguishes the four sub - sites 4@xmath1 and runs from @xmath65 to @xmath63 .
since the magnetic structure of mnsi is incommensurate , the spontaneous field differs from one particular site of the type @xmath59 to another site of the same type . in eq .
( [ eq_b_loc_zf ] ) @xmath70 represents the set of vectors defining the positions of the sites of type @xmath59 with @xmath71 , where @xmath72 defines the position of the site of type @xmath59 in the primitive cell and @xmath73 is a vector belonging to the direct crystal lattice .
the calculation of the dipolar contribution on the right - hand side of eq .
( [ eq_b_loc_zf ] ) is straightforward and will fully be determined by the knowledge of the muon stopping site and the details of the magnetic structure , such as propagation vector , value and direction of the magnetic moments , as well as helicity of the helix . on the other hand ,
the contact contribution can be written as @xmath74 where @xmath75 is the volume of one mole of mn - ions . in writing eq .
( [ eq_zf_cont ] ) , we have assumed that the contact coupling is isotropic as suggested from our tf - data .
@xmath76 is the _ local _ magnetization at the muon site @xmath70 . to compute @xmath77 ,
the first natural choice is to set @xmath78 equal to @xmath79 determined above .
this is reasonable assuming that no massive changes occur on the fermi surface when crossing @xmath2 and considering that due to the long wavelength , the close mn - neighbors around the muon , which are determining the strength of the rkky interaction , are essentially ferromagnetically aligned in the magnetic phase .
this is the only choice made for the calculation , as the _ local _ magnetization @xmath80 is also fully determined by the details of the magnetic structure and is given by @xmath81 where the volume @xmath82 is the volume occupied by the @xmath83 mn - ions involved in the sum .
the direction of the mn - moments @xmath84 is determined by the details of the magnetic structure .
for our calculations , as the rkky interaction is local , we have restricted the sum to a sphere having a radius of one lattice constant around the muon site .
the values of @xmath85 are located on the path of an ellipse for which the semimajor and semiminor axis values are @xmath86 and @xmath87 . starting from a known magnetic structure , the details for a precise computation of these values is thoroughly explained in ref . .
the present calculations where performed on a sphere containing more than @xmath88 unit cells . by setting @xmath89 , we stress again that our calculation does not contain any free parameters as the magnetic structure is taken from the literature , the muon site is determined by our tf - data and the contact coupling constant is taken _ as is _ from the tf - data . from this calculation ,
one obtains @xmath90 g ( corresponding to a muon frequency of 28.0 mhz ) , @xmath91 g ( 12.2 mhz ) and @xmath92 g ( 11.9 mhz ) .
the lower panel of fig .
[ fig_simul_zf ] shows the obtained field distributions which have been convoluted for a better visibility ( see caption ) .
the red contribution represents the sum of the field distributions for the @xmath61 , @xmath62 and @xmath63 sites whereas the blue contribution shows the field distribution expected at the @xmath65 site .
this calculation can now be directly compared to the analysis of a high statistics zf-@xmath0sr measurement performed at 5 k on a single crystal obtained according to the same procedure as the one used for the tf-@xmath0sr measurements .
we note first that the oscillatory part of the muon polarization function associated with the field distribution given by eq .
( [ eq_fd ] ) , and given by @xmath93 can not be obtained analytically .
we therefore approximate @xmath58 with a shifted overhauser distribution sr data taken at 5 k ( initial muon polarization parallel to the [ 111 ] direction .
the line represents a fit of eq .
( [ pol_func ] ) to the data using the parameters given in the text ( for a better visibility a time - binning corresponding to 5 ns was choosen , whereas the fits were performed with a time - binning of 1.25 ns ) . ]
@xmath94 where @xmath95 and @xmath96 .
the upper panel of fig .
[ fig_approx_bessel ] shows the difference between the exact and approximated field distribution , which is symmetrical with respect to the singularities , i.e. some weight is transfered from the upper cutoff to the lower cutoff field .
the lower panel of fig .
[ fig_approx_bessel ] shows the oscillatory part of the muon polarization function obtained for both field distributions . for the exact field distribution a numerical calculation
was performed using eq .
( [ p_osc ] ) , whereas for the approximated field distribution the muon polarization function can be obtained analytically and is given by @xmath97 where @xmath98 is a bessel function of the first kind .
we see that the function @xmath99 catches the essential features of @xmath100 and constitutes a good approximation . having this in mind ,
we can now write a function which will be fitted to the zf experimental data , i.e. sr data obtained at 5 k and shown on fig . [ fig_zf_musr_spectra_5k ] .
the blue and red components represent respectively the first and second component of the right - hand side of eq ( [ pol_func ] ) .
lower panel : computed field distributions for the @xmath65-site ( blue ) and for the @xmath61 , @xmath62 and @xmath63 sites ( red ) . for a better visibility , these field distributions have been slightly folded by lorentzian distributions with widths ( fwhm ) of 5 and 20 g , respectively . ]
@xmath101 where the first component is associated to the site @xmath65 and the second to the sites @xmath61 , @xmath62 and @xmath63 .
the depolarization rates reflect as usual any static or dynamical effects , the discussion of which goes beyond the scope of this study .
figure [ fig_zf_musr_spectra_5k ] shows the zf-@xmath0sr spectra with the fitted polarization function . from the fitted parameters one obtains @xmath102 g ( 12.34 mhz ) , @xmath103 g ( 12.99 mhz ) and @xmath104 g ( 28.06 mhz ) .
these values agree very well with our calculations performed without free parameters .
in addition , both phase parameters @xmath105 and @xmath106 were found to be compatible with zero , as expected .
the depolarization rates are found to have the values @xmath107 and @xmath108 .
we note that the small difference between the values of @xmath109 and @xmath110 ( which is also reflected in our calculations ) perfectly explains that solely two peaks were invoked in former @xmath0sr data realized with much less statistics .
finally an interesting point is the fact that @xmath111 in both the calculations and the experimental data .
this is remarkable as the position of @xmath112 , with respect to @xmath113 , depends on the helicity of the incommensurate helix and is located above @xmath113 only for left - handed helicity .
it is therefore tempting to take the zf-@xmath0sr data as a late confirmation of the helix helicity .
we note also that a close look at the fast fourier transform reveals as expected a very slight divergence near the cutoff fields between the data and the shifted overhauser distribution ( see the upper panel of fig .
[ fig_simul_zf ] and compare the data points to the red distribution ) .
we therefore conclude here that from the knowledge of the muon site determined by our tf - data , one can utterly explain the characteristics of the zf-@xmath0sr data . for completeness
, we compare now our precise experimental determination of the muon site with _ ab initio _ calculations based on the density - functional theory ( dft ) .
such calculations have recently been shown to accurately reproduce the muon sites observed in different materials as wide - gap semiconductors , insulating systems or cuprates and iron - based high-@xmath34 superconductors @xcite .
the calculations were performed at the university of parma and we like to stress that they have been performed in a _ blind modus _ ,
i.e. without any prior knowledge of the muon site deduced from the tf-@xmath0sr data . .
the yellow regions define isosurfaces with an energy set to @xmath114 with @xmath115 being the ground state energy obtained from the solution of the schrdinger equation for the muon in the electrostatic potential .
the equivalent muon stopping sites determined by tf-@xmath0sr are also indicated and are enclosed in the minima regions .
note that compared to fig .
[ structure ] , we also report the muon sites located just outside of the primary unit cell , which are enclosed in minima regions extending into the primary unit cell . ] to analyze the electronic structure of mnsi we use the fp - lapw approach as implemented in the elk code @xcite .
the generalized gradient approximation ( gga ) as formulated by perdew , burke and ernzerhof is used to approximate the exchange and correlation potential@xcite .
the cutoff for the plane wave expansion in the interstitial region is set to @xmath116 = 9 ( @xmath117 being the smallest muffin - tin radius ) .
the reciprocal space is sampled on a @xmath118 monkhorst - pack@xcite grid .
the experimental atomic positions and lattice constant are used .
a ferromagnetic ground state is considered and the resulting magnetic moment on the mn atoms is approximately 1 @xmath119 .
indeed mnsi in a well known case in which the mean field approximation leads to a overestimation of the magnetic moments @xcite .
the muon site is identified from the minima of the electrostatic potential obtained from the ground state electronic density of the unperturbed material .
there are four equivalent minima in the unit cell , as show in fig .
[ fig_dft ] , which correspond to the @xmath20 wyckoff position with fractional coordinates ( 0.523,0.523,0.523 ) .
this coarse estimation nicely agrees with the experimentally evaluated site .
the standard procedure@xcite for the dft site assignment further requires the analysis of the perturbation introduced by the muon in the vicinity of its embedding site .
this is done by analyzing the electronic and crystallographic modifications introduced by the charged impurity .
the muon site is eventually validated by considering the spread of its wave - function .
these steps are computationally demanding and outside the scope of the present work .
however preliminary indications suggest that the inclusion of the muon in a supercell produces an even closer agreement with the experimental value , providing a 4@xmath1 muon stopping site at the coordinates ( 0.538,0.538,0.538 ) .
we conclude that the observed angular dependence of the tf-@xmath0sr signals ( except for a very slight dependence due to the demagnetization factor ) can be beyond any doubt ascribed by a muon sitting at a 4@xmath1 wyckoff position and detecting the dipolar field produced by the mn moments .
consequently the @xmath0sr response of paramagnetic mnsi under the application of a magnetic field can be fully understood without invoking a hypothetical magnetic polaron state .
in addition we have shown that the knowledge of the muon stopping site provides a clear understanding of the zf-@xmath0sr data and that the computed field distribution perfectly agrees with the measured one . moreover , _
blind modus ab initio _ dft calculations provide a 4@xmath1 muon stopping which is in very close agreement with the site determined by the experimental @xmath0sr data .
d.a . acknowledges partial financial support from the romanian uefiscdi project no .
pn - ii - id - pce-2011 - 3 - 0583 ( 85/2011 ) .
, p.b . and f.b .
acknowledge partial support of prin grant 2012x3yfz2_004 .
part of this work was performed at the swiss muon source , paul scherrer institut , villigen , switzerland .
e. bauer and m. sigrist , series : lecture notes in physics , vol .
847 ( springer ) ( 2012 ) .
i.e. dzialoshinskii , sov .
jetp 5 , 1259 ( 1957 ) .
t. moriya , phys .
120 , 91 ( 1960 ) .
v.p . mineev and m. sigrist , in _ non - centrosymmetric superconductors : introduction and overview _ , edited by e. bauer and m. sigrist ( springer , berlin , 2012 ) , p. 129 .
y. ishikawa _ et al .
_ , j. phys .
19 , 525 ( 1976 ) .
y. ishikawa _ et al .
b 16 , 4956 ( 1977 ) .
m. ishida , _ et al .
_ , j. phys
54 , 2975 ( 1985 ) .
thompson , z. fisk , and g. g. lonzarich , physica b 161 , 317 ( 1989 ) .
y. ishikawa _
et al . _ , j. phys .
japan , 53 , 2726 ( 1984 ) .
s. mhlbauer _
_ , science 323 , 915 ( 2009 ) .
g. goll , _ unconventional superconductors _ ( springer , berlin , 2006 ) .
s.s . saxena _
et al . _ , nature , 406 , 587 ( 2000 ) .
et al . _ ,
41 , 1743 ( 1978 ) .
r. kadono _ et al .
_ , phys . rev .
b 42 , 6515 ( 1990 ) .
a. yaouanc _
et al._,j .
matter 17 , l129 ( 2005 ) .
et al . _ , nature physics 3 , 29 ( 2007 ) .
d. andreica _ et al .
_ , phys . rev .
b 81 , 060412(r ) ( 2010 ) .
m. takigawa _
inter . 8 , 773 ( 1981 ) .
v. storchak _
et al . _ , phys . rev .
b 83 , 140404(r ) ( 2011 ) .
a. suter and b.m .
wojek , physics procedia 30 , 69 ( 2012 ) .
a. yaouanc and p. dalmas de rotier , _ muon spin rotation , relaxation and resonance : applications to condensed matter _ ( oxford university press , oxford , 2011 ) .
104 , 115 ( 1997 ) .
r. feyerherm _ et al .
_ , z. phys .
b 99 , 3 ( 1995 ) . note that in ref . and the eq .
( [ adipgeneral ] ) is expressed with the terms containing the polynomial @xmath120 as being positive .
the difference arises from the definition of the associated legendre polynomial .
we adopt here the common expression , i.e. @xmath121 .
_ , rc mat .
bull . 7 , 1431 ( 1972 ) . s.m .
stishov _ et al .
b 76 , 052405 ( 2007 ) .
stishov , a.e .
petrova , physics - uspekhi 54 , 1117 ( 2012 ) .
bear in mind that , strictly speaking , we should actually only consider the contribution of the @xmath42 electrons to the total susceptibility .
however , the fact that the total susceptibility perfectly follows a @xmath122-temperature dependence @xcite is a strong indication that the response of the local moments will dominate the susceptibility upon decreasing the temperature .
as our measurements were performed at rather low temperature we can safely assume that @xmath123 . in any case ,
as explained later in the text , an even large uncertainty on the value of the susceptibility will only very weakly affect the position of the obtained muon site . however , the conclusion that the muon is stopped at a @xmath20 position is not affected by a variation of @xmath31 .
et al . _ ,
b 43 , 3284 ( 1991 ) .
even though the wavelength of the magnetic structure is long , the system is antiferromagnetic and the lorentz field is zero , since the magnetization integrated outside of the lorentz sphere vanishes .
a. amato , rev .
mod . phys .
69 , 1119 ( 1997 ) .
d. andreica , phd thesis , eth zurich , nr .
14170 ( 2001 ) .
a. schenck
_ et al . _ ,
b 65 , 024444 ( 2001 )
. f. bernardini _
et al . _ ,
b 87 , 115148 ( 2013 ) . j.s .
et al . _ , phys .
b 87 , 121108(r ) ( 2013 ) .
s.j . blundell _
et al . _ ,
b 88 , 064423 ( 2013 ) .
et al . _ ,
88 , 068510 ( 2013 ) .
elk code , version 2.2.10 .
perdew _ et al .
77 , 3865 ( 1996 ) .
monkhorst and j.d .
pack , phys .
b 13 , 5188 ( 1976 ) . t. jeong and w.e .
pickett , phys .
b 70,075114 ( 2004 ) r.d . collyer and d.a .
browne , physica b : condensed matter 403 , 1420 ( 2008 ) . | transverse - field muon - spin rotation ( @xmath0sr ) experiments were performed on a single crystal sample of the non - centrosymmetric system mnsi .
the observed angular dependence of the muon precession frequencies matches perfectly the one of the mn - dipolar fields acting on the muons stopping at a 4@xmath1 position of the crystallographic structure .
the data provide a precise determination of the magnetic dipolar tensor .
in addition , we have calculated the shape of the field distribution expected below the magnetic transition temperature @xmath2 at the 4@xmath1 muon - site when no external magnetic field is applied .
we show that this field distribution is consistent with the one reported by zero - field @xmath0sr studies .
finally , we present _ ab initio _ calculations based on the density - functional theory which confirm the position of the muon stopping site inferred from transverse - field @xmath0sr . in view of the presented evidence
we conclude that the @xmath0sr response of mnsi can be perfectly and fully understood without invoking a hypothetical magnetic polaron state . |
in a recent letter @xcite it was pointed out that as a consequence of helicity - flip bremsstrahlung , electrons in the radiative decay @xmath0 are not purely left - handed , even in the limit @xmath5 . the decay width into right - handed electrons was shown to be @xmath6 , where @xmath7 .
the energy spectrum of these wrong - helicity electrons was calculated , as also that of the photons accompanying them .
these results were obtained using the helicity - flip splitting function @xmath8 introduced by falk and sehgal @xcite .
the appearance of right - handed electrons in @xmath9-decay is , at first sight , surprising since it goes against the conventional wisdom based on the v - a structure of weak interaction and the common assumption of helicity - conservation in massless qed
. however persistence of helicity - flip bremsstrahlung in the @xmath5 limit has been established in several calculations of radiative processes @xcite , going back to the seminal paper of lee and nauenberg @xcite . in all cases ,
the origin of the effect is the characteristic behaviour of helicity - flip bremsstrahlung at small angles @xmath10 ^ 2 } \,,\ ] ] leading to a non - zero integrated rate that survives the limit @xmath11 . in the present paper ,
we derive the helicity - dependent spectrum of the decay @xmath0 from first principles , without neglecting the electron mass .
we obtain the longitudinal polarization @xmath12 of the electron as a function of the electron energy , photon energy and the angle between the electron and photon .
we then consider the limit @xmath5 to demonstrate that the deviation from @xmath13 is present even in the chiral limit of qed .
the results are compared with those obtained in the equivalent particle approach based on the helicity - flip fragmentation function , thus providing a confirmation of the statements in ref.@xcite .
the matrix element for the decay @xmath0 may be obtained from the feynman diagrams shown in fig.([feyndia ] ) .
( 370,120 ) ( 78,50)(149,50 ) ( 78,50)(128,0 ) ( 0,50)(78,50 ) ( 78,50)(128,100 ) ( 93,65)(118,117)34 ( 17,33)[lb ] ( 93,93)[lb ] ( 129,85)[lb ] ( 154,45)[lb ] ( 129,8)[lb ] ( 178,45)[lb ] ( 278,50)(349,50 ) ( 278,50)(328,0 ) ( 200,50)(278,50 ) ( 278,50)(328,100 ) ( 258,50)(283,102)34 ( 217,33)[lb ] ( 258,78)[lb ] ( 329,85)[lb ] ( 354,45)[lb ] ( 329,8)[lb ] adopting the notation of ref.@xcite , we obtain the square of the matrix element , summed over the photon polarization , taking the muon to be unpolarized .
the spectrum is derived in terms of the invariants @xmath14 and the mass ratio @xmath15 . in the rest frame of the muon , @xmath16 \,,\ ] ] @xmath17 being the energies of the electron , photon and @xmath18 the angle between them .
the boundary of phase space is determined by the condition @xmath19 , and @xmath20 , @xmath21 being the `` missing mass '' of the neutrino pair .
this leads to the condition @xmath22 where @xmath23 equivalently , the limits of phase space may be expressed as @xmath24 where @xmath25 the differential decay width for @xmath0 in the variables @xmath26 , @xmath27 and @xmath18 is then given by @xmath28 where r and l denote right - handed and left - handed electrons , corresponding to the polarization vectors @xmath29 the functions @xmath30 and @xmath31 are given by @xmath32 @xmath33 adding the expressions for @xmath34 and @xmath35 , we obtain the spectrum for unpolarized electrons , which depends on the function @xmath30 only @xcite .
the specific effects of electron helicity are contained in the function @xmath31 .
summing over the electron helicity , the differential decay rate of @xmath0 is @xmath36 from this we can derive the following spectra .
@xmath37 \big\ } \\ &
\phantom{=\gamma_{0}\,\frac{\alpha}{6\,\pi}\,\frac{1}{y}\,\big(}+ 12\,{{\mathcal{l}}}\ , \big\ { -4\,x^3 + x^2\,(6 - 8\,y ) + 6\,{\lambda}^4\,y - 6\,x\,(-1 + y)\,y + ( 3 - 2\,y)\,y^2 \\ & \phantom{=\gamma_{0}\,\frac{\alpha}{6\,\pi}\,\frac{1}{y}\,\big(+ 12\,{{\mathcal{l}}}\,\big\ { } + { \lambda}^2\,[6\,x^2 + ( 6 - 5\,y)\,y - 2\,x\,(4 + y ) ] \big\ } \big)\ , , \end{split}\ ] ] where we have defined @xmath38 the photon energy spectrum @xmath39 is given explicitly in eq.([ps - rl ] ) of the appendix the leading term for small @xmath27 ( small photon energies ) is @xmath40 \,.\ ] ] in the limit of small @xmath41 , this reduces to the approximate form @xmath42 \,,\ ] ] which agrees with the old calculation of eckstein and pratt @xcite and kinoshita and sirlin @xcite , in which terms of order @xmath43 and higher powers were neglected .
the integrated rate for @xmath44 is @xmath45 \,,\ ] ] which agrees with ref .
@xcite , except in the constant ( non - logarithmic ) term , which was @xmath46 in the old calculation . the exact result for the electron energy spectrum @xmath47 is given in eq.([xi2-u-1 ] ) of the appendix .
the normalized electron energy spectrum is shown in fig.([xspectrum1]a ) .
1truecm in the limit @xmath4 we have for the @xmath26-spectrum @xmath48 \\ & \phantom{\big\{}+ 12\,x^2\,(-3 + 2\,x)\ , \log \big ( \frac{1 - x}{y_0 } \big ) \big\ } \\
& \times\ , \big [ ( 1 - y_0)\,(7 - 3\,y_0 + 3\,{y_0}^2 - { y_0}^3 ) + 6\,\log ( y_0 ) \big]^{-1 } \ , . \end{split}\ ] ] this is shown in fig.([xspectrum1]b ) . in the limit @xmath49
we reproduce the spectrum for non - radiative muon decay @xmath50
the spectra when the final electron is polarized in the state r or l are obtained using eq.([dgamma1 ] ) .
we will emphasize the case r , which is the one which behaves in an unexpected way in the @xmath51 limit @xcite .
the corresponding spectra for the state l are obtained by subtraction from the unpolarized result l+r discussed in sec.[unpolspectra ] .
@xmath52 \\ & \phantom{\big\{}+ { { \mathcal{a}}}\,({{\mathcal{a}}}- x)\ , \big [ -72\,x + 48\,x^2 - 72\,y + 78\,x\,y + 24\,x^2\,y + 33\,x\,y^2 \\ & \phantom{\big\{+ { { \mathcal{a}}}\,({{\mathcal{a}}}- x)\,\big [ } - 10\,x^2\,y^2 + 12\,y^3 - 6\,x\,y^3 + 4\,x^2\,y^3 \big ] \\ & \phantom{\big\{}+{{\mathcal{l}}}\ , \big [ -12\,({{\mathcal{a}}}- x)\,(-3 + 2\,x + 2\,y)\,(2\,x^2 + 2\,x\,y + y^2 ) \\ & \phantom{\big\{+{{\mathcal{l}}}\,\big [ } + { \lambda}^4\ , ( 96\,x - 192\,y + 72\,{{\mathcal{a}}}\,y - 24\,x\,y + 48\,y^2 ) \\ & \phantom{\big\{+{{\mathcal{l}}}\,\big [ } + { \lambda}^2\,(288\,x - 96\,{{\mathcal{a}}}\,x - 192\,x^2 + 72\,{{\mathcal{a}}}\,x^2 - 24\,x^3 + 72\,{{\mathcal{a}}}\,y - 264\,x\,y \\ & \phantom{\big\{+{{\mathcal{l}}}\,\big[+ { \lambda}^2\ , ( } - 24\,{{\mathcal{a}}}\,x\,y + 72\,x^2\,y - 48\,y^2 - 60\,{{\mathcal{a}}}\,y^2 + 84\,x\,y^2 + 48\,y^3 ) \big ] \big\ } \ , .
\\ \end{split}\ ] ] in the limit @xmath4 we obtain @xmath53 this limiting result coincides with what one obtains on the basis of the helicity - flip splitting function .
denoting the electron spectrum in the non - radiative decay @xmath54 , the spectrum after helicity - flip bremsstrahlung is @xmath55 noting that @xmath56 , and recalling that @xmath57 we obtain @xmath58 exactly as in eq.([limit1 ] ) .
of particular interest is the photon energy spectrum associated with the right - handed polarization state of the electron , which is naively forbidden in the @xmath4 limit .
the corresponding spectrum for l - electrons may be obtained by subtraction from eq.([ps - rl ] ) . the exact result for @xmath59
is stated in eq.([photon - spectrum - r - eq ] ) . in the limit @xmath60 one
obtains @xmath61 which is the result derived in @xcite on the basis of the helicity - flip fragmentation function @xmath62 . the spectrum for @xmath63 is plotted in fig.([psrplot1 ] ) .
1truecm ( 8,4.96 ) ( 0,0 ) ( 4,3.8)a ( 5.3,4)b ( 4.2,-0.2)y the striking feature is the appearance of a divergence at @xmath64 .
this is connected with the fact that for @xmath65 , helicity is not a good quantum number . as a consequence , ordinary ( non - flip )
bremsstrahlung , diverging as @xmath66 at small @xmath27 , is partly transmitted to the final state with electron polarization @xmath67 .
this contribution vanishes in the limit @xmath4 , leaving behind the anomalous ( helicity - flip ) contribution in eq.([psrlimit ] ) .
this anomalous bremsstrahlung integrates to a decay width @xmath68 which is the contribution of right - handed electrons to the muon decay width in the @xmath4 limit @xcite . the exact expression for @xmath69 and the corresponding left - handed rate @xmath70
are plotted in fig.([widthplot1 ] ) . from these
, we obtain the longitudinal polarization 1truecm ( 10,5 ) ( 0,0 ) ( 5,-0.2)@xmath71 ( 1.2,2.2)r @xmath72 ( 2.3,3.5)l @xmath73 plotted in fig.([polplot2 ] ) .
1truecm ( 10,6 ) ( 0,0 ) ( 5.2,-0.2)@xmath71 ( 7.7,2.5)a ( 8.7,1)b this figure demonstrates the presence of a significant right - handed electron component in radiative muon decay , which becomes as probable as the left - handed component at high photon energies . to illustrate the approach to the chiral limit @xmath4
, we have also shown in fig.([polplot2 ] ) the polarization of the electron in the decay @xmath74 . in the chiral limit
@xmath4 the longitudinal polarization goes from @xmath13 to @xmath75 like a step - function at @xmath76 .
the formula for the electron energy spectrum @xmath77 for right - handed electrons is given in eq.([psr-1 ] ) , and may be compared with the unpolarized spectrum @xmath47 in eq.([xi2-u-1 ] ) we have verified that in the limit @xmath4 , the right - handed spectrum takes the form @xmath78 \,.\ ] ] this is exactly the result derived in ref.@xcite on the basis of the fragmentation function @xmath62 .
a question of interest is the behaviour of helicity - flip and helicity - non - flip bremsstrahlung in the collinear region , @xmath79 , where @xmath80 is the angle between the photon and electron in the final state . to this end , we have calculated exactly the distribution @xmath81 .
the analytic results are , however , too lengthy to be written down here . the left- and right - handed spectra for @xmath82 , in the small angle region ,
are plotted in fig.([ctplot1 ] ) .
1truecm ( 10,6.2 ) ( 0,0 ) ( 6,3.7)l ( 6,1.2)r ( 4.8,-0.4)@xmath18 the longitudinal polarization of the electron as a function of @xmath80 is shown in fig.([ctpolplot2 ] ) , and , changes from the v - a value @xmath83 to the value @xmath84 in the forward direction .
1truecm ( 10,6.2 ) ( 0,0 ) ( 4.9,-0.3)@xmath18 from the exact expressions we have derived the behaviour in the chiral limit @xmath4 . for @xmath85 and @xmath4 , we find @xmath86 \ , . \end{split}\ ] ] note that , in the chiral limit , the @xmath87 behaviour of @xmath88 refers to the opening angle of the electron and photon in the final state .
the corrsponding @xmath89 behaviour of the helicity - flip scattering cross section ( eq.([cross - section-1 ] ) ) refers to the angle between the photon and the incident electron .
we have also derived the value of @xmath90 in the forward direction ( @xmath91 ) , for small @xmath41 : @xmath92 \big ( \frac{{{\rm{d}}}\gamma}{{{\rm{d}}}\cos(\theta ) } \big)_{\rm{l } } \bigg|_{\theta = 0 } & = \gamma_0\,\frac{\alpha}{\pi}\,\frac{1}{120}\,(1 - y_0)^4\,(1 + 4 y_0 ) \ , .
\end{split}\ ] ] we have found that the angular distribution @xmath93 connected with helicity - flip bremsstrahlung can be simulated by the following ansatz for the differential decay spectrum ( where the label `` app '' connotes `` approximate '' ) : @xmath94 when integrated over @xmath95 , this spectrum , in the limit @xmath4 , yields the dalitz pair distribution given in eq.([limit1 ] ) . after integration over @xmath26 and @xmath27 we obtain an angular distribution @xmath96 which agrees with the exact calculation over a wide range of angles . this formula is particularly simple for @xmath97 : @xmath98\ , \frac{1}{\sin^2 ( \theta ) } \,,\ ] ] which agrees with the small angle result given in eq.([ct - lim - r-1 ] ) .
similarly , in the forward direction ( @xmath91 ) , for small values of @xmath41 , the approximate representation eq.([app-1 ] ) reproduces the result given in eq.([coll - lim - r-1 ] ) .
the purpose of this paper was to examine the dependence of the radiative decay @xmath0 on the helicity of the final electron , in order to demonstrate the presence of right - handed electrons , even in the chiral limit @xmath5 .
this non - intuitive feature is a consequence of helicity - flip bremsstrahlung in qed , and was discussed in @xcite with the help of the helicity - flip splitting function derived in @xcite .
we have now calculated the complete helicity - dependent spectrum of this decay from first principles , without neglecting the electron mass , and have shown that the results of @xcite are reproduced in the limit @xmath4 .
the presence of right - handed electrons manifests itself in the longitudinal polarization @xmath99 which deviates from the naive v - a value @xmath13 even in the limit @xmath4 .
the dependence of the polarization on the photon energy cut @xmath71 , fig.([polplot2 ] ) , shows that right - handed electrons occur predominantly with high energy photons .
a characteristic signature of wrong - helicity electrons is also present in the opening - angle distribution @xmath100 , where the collinear region @xmath101 is dominated by right - handed electrons .
the fact that helicity - flip bremsstrahlung in the chiral limit @xmath5 can be exactly reproduced by the simple and universal splitting function @xmath102 in many different processes , is suggestive of a relationship to the chiral anomaly in qed @xcite .
@xmath103 \\ & \phantom{\gamma_0\,\frac{\alpha}{36\,\pi}\,\frac{1}{y}\,\big\ { } + \big [ 24\,(-3 + 2\,y - 4\,y^2 + 2\,y^3)\,(1 - y ) + 72\,\lambda^2\,y\,(-13 + 19\,y - 7\,y^2)\,(1 - y)^{-1 } \\ & \phantom{\gamma_0\,\frac{\alpha}{36\,\pi}\,\frac{1}{y}\,\big\{\big [ } + 72\,\lambda^4\,(12 - 27\,y + 18\,y^2 - 5\,y^3)\,(1 - y)^{-1 } + 24\,\lambda^6\,y\,(5 + 15\,y - 18\,y^2)\,(1 - y)^{-3 } \\ & \phantom{\gamma_0\,\frac{\alpha}{36\,\pi}\,\frac{1}{y}\,\big\{\big [ } + 72\,\lambda^8\,(-1 + 2\,y)\,(1 - y)^{-4 } \big]\ , \log \big ( \frac{\lambda}{1 - y } \big ) \big\ } \ , .
\end{split}\ ] ] where @xmath105 , and @xmath106 \ , , \\ k_2 \,:=\,\ , & \frac{1}{3}\,u\,(1 - y_0 - u)\ , \big [ u^2\ , \big ( 5 + 17\,u - 34\,u^2 + 5\,y_0 - 14\,u\,y_0 - 4\,{y_0}^2
\big ) + 3\,{\lambda}^2\,u\ , \big ( 6 - 19\,u
+ u^2 - 6\,y_0 - 5\,u\,y_0 \big ) \\ & \phantom{\frac{1}{3}\,u\,(1 - y_0 - u)\,\big[}+ 12\,{\lambda}^4\ , \big ( -4 - u + 3\,u^2 \big ) \big ] \ , , \\ k_3 \,:=\,\ , & 4\,({\lambda}^2 - u^2)\ , \big [ u^3\,(3 - 2\,u ) + { \lambda}^2\,u\,(3 - 8\,u + 3\,u^2 ) + { \lambda}^4\,(-2 + 3\,u ) \big ] \ , , \\ k_4 \,:=\,\ , & 4\,({\lambda}^2 + u^2)\ , \big [ u^3\,(3 - 2\,u ) + { \lambda}^2\,u\,(3 - 8\,u + 3\,u^2 ) + { \lambda}^4\,(-2 + 3\,u ) \big ] \ , , \end{split}\ ] ] where @xmath110 \\ & \phantom{\frac{1}{36}\,\frac{1}{u^3}\,(1 - y_0 - u)\,\big\ { } + \lambda^2\,u^2\ , \big [ 264 - 168\,y_0 - 24\,{y_0}^2 + u\,(261 - 15\,y_0 - 12\,{y_0}^2 ) \\ & \phantom{\frac{1}{36}\,\frac{1}{u^3}\,(1 - y_0 - u)\,\big\ { + \lambda^2\,u^2\ , } + u^2\,(-155 - 26\,y_0 + 4\,{y_0}^2 ) + u^3\,(-5
+ 5\,y_0 ) + u^4\,(-5 ) \big ] \\ & \phantom{\frac{1}{36}\,\frac{1}{u^3}\,(1 - y_0 - u)\,\big\ { } + \lambda^4\,u\ , \big [ -555 - 87\,y_0 + 12\,{y_0}^2 + u\,(199 - 32\,y_0 + 4\,{y_0}^2 ) \\ & \phantom{\frac{1}{36}\,\frac{1}{u^3}\,(1 - y_0 - u)\,\big\ { + \lambda^4\,u\ , } + u^2\,(73 + 5\,y_0 ) + u^3\,(-5 ) \big ] \\ & \phantom{\frac{1}{36}\,\frac{1}{u^3}\,(1 - y_0 - u)\,\big\ { } + \lambda^6\ , \big [ -122 + 22\,y_0 - 8\,{y_0}^2 + 10\,u\,(4 - y_0 ) + 10\,u^2 \big ] \big\ } \ , , \\ j_2 \,:=\,\ , & \frac{1}{3}\,\frac{1}{u^2}\,\frac{{\lambda}^2}{u^2 - { \lambda}^2}\ , \big\ { u^2\,(1 - y_0 - u)\ , \big [ -5 - 5\,y_0 + 4\,{y_0}^2 + u\,(-19 + 12\,y_0 + 4\,{y_0}^2 ) + u^2\,(-13 - y_0 ) + 13\,u^3 \big ] \\ & \phantom{\frac{1}{3}\,\frac{1}{u^2}\,\frac{{\lambda}^2}{u^2 - { \lambda}^2}\,\big\ { } + 6\,{\lambda}^2\,u\,(1 - y_0 - u)\ , \big [ -3 + 3\,y_0 + u\,(2 + 3\,y_0 ) + u^2\,(-4 + y_0 ) + u^3 \big ] \\ & \phantom{\frac{1}{3}\,\frac{1}{u^2}\,\frac{{\lambda}^2}{u^2 - { \lambda}^2}\,\big\ { } + 24\,{\lambda}^4\,(2 - u)(1 + u)\,(1 - y_0 - u ) \big\ } \ , , \\
j_3 \,:=\,\ , & 4\,\frac{1}{u^3}\,{\lambda}^2\,(1 - u)\ , ( { \lambda}^2 - u^2)\,(-2\,{\lambda}^2 + 3\,u - u^2 ) \ , , \\ j_4 \,:=\,\ , & 4\,\frac{1}{u^3}\,{\lambda}^2\,(1 - u)\ , ( { \lambda}^2 + u^2)\,(-2\,{\lambda}^2 + 3\,u - u^2 ) \ , . \end{split}\ ] ] l. m. sehgal , phys .
b * 569 * , 25 ( 2003 ) .
b. falk and l. m. sehgal , phys .
b * 325 * , 509 ( 1994 ) .
h. f. contopanagos and m. b. einhorn , nucl .
b * 377 * , 20 ( 1992 ) .
s. jadach , j. h. khn , r. g. stuart and z. was , z. phys .
c * 38 * , 609 ( 1988 ) [ erratum - ibid .
c * 45 * , 528 ( 1990 ) ] .
l. trentadue and m. verbeni , phys .
b * 478 * , 137 ( 2000 ) and nucl .
b * 583 * , 307 ( 2000 ) ; a. v. smilga , comments nucl . part .
phys . * 20 * , 69 ( 1991 ) .
s. groote , j. g. krner and m. m. tung , z. phys .
c * 70 * , 281 ( 1996 ) .
w. e. fischer and f. scheck , nucl .
b * 83 * , 25 ( 1974 ) .
t. d. lee and m. nauenberg , phys .
rev . * 133 * , b1549 ( 1964 ) .
m. fischer , s. groote , j. g. krner and m. c. mauser , phys .
d * 67 * , 113008 ( 2003 ) .
a. fischer , t. kurosu and f. savatier , phys .
d * 49 * , 3426 ( 1994 ) .
s. g. eckstein and r. r. pratt , ann .
* 8 * , 297 ( 1959 ) .
t. kinoshita and a. sirlin , phys .
rev . * 113 * , 1652 ( 1959 ) .
a. d. dolgov and v. i. zakharov , nucl .
b * 27 * , 525 ( 1971 ) ; k. huang , _
quarks , leptons and gauge fields _ ( world scientific , singapore , 1992 ) ; j. horejsi , czech .
j. phys .
* 42 * , 241 ( 1992 ) .
a. freund and l. m. sehgal , phys .
b * 341 * , 90 ( 1994 ) .
r. d. carlitz , j. c. collins and a. h. mueller , phys .
b * 214 * , 229 ( 1988 ) . | we have studied in detail the spectrum of the decay @xmath0 as a function of the electron helicity , and verify the prediction @xcite that a significant fraction of the electrons in this decay is right - handed .
these `` wrong - helicity '' electrons persist in the limit @xmath1 , and are connected with helicity - flip bremsstrahlung in qed .
the longitudinal polarization of the electron is calculated as a function of the photon and electron energy , and deviates systematically from the naive v - a prediction @xmath2 .
the right - handed component is concentrated in the collinear region @xmath3 . in the limit @xmath4
, we reproduce the results obtained in @xcite using the helicity - flip splitting function introduced by falk and sehgal . |
the need for rapid implementation of high performance , robust , and portable finite element methods has led to approaches based on automated code generation .
this has been proven successful in the context of the fenics @xcite and firedrake @xcite projects . in these frameworks ,
the weak variational form of a problem is expressed in a high level mathematical syntax by means of the domain - specific language ufl @xcite .
this mathematical specification is used by a domain - specific compiler , known as a form compiler , to generate low - level c or c++ code for the integration over a single element of the computational mesh of the variational problem s left and right hand side operators .
the code for assembly operators must be carefully optimized : as the complexity of a variational form increases , in terms of number of derivatives , pre - multiplying functions , or polynomial order of the chosen function spaces , the operation count increases , with the result that assembly often accounts for a significant fraction of the overall runtime . as demonstrated by the substantial body of research on the topic , automating the generation of such high performance implementations poses several challenges .
this is a result of the complexity inherent in the mathematical expressions involved in the numerical integration , which varies from problem to problem , and the particular structure of the loop nests enclosing the integrals .
general - purpose compilers , such as those by _
gnu _ and _ intel _ , fail to exploit the structure inherent in the expressions , thus producing sub - optimal code ( i.e. , code which performs more floating - point operations , or `` flops '' , than necessary ; we show this in section [ sec : perf - results ] ) .
research compilers , for instance those based on polyhedral analysis of loop nests , such as pluto @xcite , focus on parallelization and optimization for cache locality , treating issues orthogonal to the question of minimising flops .
the lack of suitable third - party tools has led to the development of a number of domain - specific code transformation ( or synthesizer ) systems .
show how automated code generation can be leveraged to introduce optimizations that a user should not be expected to write `` by hand '' . and
employ mathematical reformulations of finite element integration with the aim of minimizing the operation count .
in , the effects and the interplay of generalized code motion and a set of low level optimizations are analysed .
it is also worth mentioning two new new form compilers , uflacs @xcite and tsfc @xcite , which particularly target the compilation time challenges of the more complex variational forms .
the performance evaluation in section [ sec : perf - results ] includes most of these systems .
however , in spite of such a considerable research effort , there is still no answer to one fundamental question : can we automatically generate an implementation of a form which is optimal in the number of flops executed ? in this paper , we formulate an approach that solves this problem for a particular class of forms and provides very good approximations in all other cases . in particular , we will define `` local optimality '' , which relates operation count with inner loops . in summary ,
our contributions are as follows : * we formalize the class of finite element integration loop nests and we build the space of legal transformations impacting their operation count . * we provide an algorithm to select points in the transformation space .
the algorithm uses a cost model to : ( i ) understand whether a transformation reduces or increases the operation count ; ( ii ) choose between different ( non - composable ) transformations .
* we demonstrate that our approach systematically leads to a local optimum .
we also explain under what conditions of the input problem global optimality is achieved . *
we integrate our approach with a compiler , coffee , which is in use in the firedrake framework . *
we experimentally evaluate using a broader suite of forms , discretizations , and code generation systems than has been used in prior research .
this is essential to demonstrate that our optimality model holds in practice .
in addition , in order to place coffee on the same level as other code generation systems from the viewpoint of low level optimization ( which is essential for a fair performance comparison ) : * we introduce a transformation based on symbolic execution that allows irrelevant floating point operations to be skipped ( for example those involving zero - valued quantities ) . after reviewing basic concepts in finite element integration , in section [ sec : lnopt ]
we introduce a set of definitions mapping mathematical properties to the level of loop nests .
this step is an essential precursor to the definition of the two algorithms sharing elimination ( section [ sec : sharing - elimination ] ) and pre - evaluation ( section [ sec : pre - evaluation ] ) through which we construct the space of legal transformations .
the main transformation algorithm in section [ sec : optimal - synthesis ] delivers the local optimality claim by using a cost model to coordinate the application of sharing elimination and pre - evaluation .
we elaborate on the correctness of the methodology in section [ sec : proof ] .
the numerical experiments are showed in section [ sec : perf - results ] .
we conclude discussing the limitations of the algorithms presented and future work .
we review finite element integration using the same notation and examples adopted in and .
consider the weak formulation of a linear variational problem : @xmath1 where @xmath2 and @xmath3 are , respectively , a bilinear and a linear form .
the set of _ trial _ functions @xmath4 and the set of _ test _ functions @xmath5 are suitable discrete function spaces . for simplicity , we assume @xmath6 .
let @xmath7 be the set of basis functions spanning @xmath4 .
the unknown solution @xmath8 can be approximated as a linear combination of the basis functions @xmath7 . from the solution of the following linear system it is possible to determine a set of coefficients to express @xmath8 : @xmath9 in which @xmath10 and @xmath11 discretize @xmath2 and @xmath3 respectively : @xmath12 the matrix @xmath10 and the vector @xmath11 are assembled and subsequently used to solve the linear system through ( typically ) an iterative method .
we focus on the assembly phase , which is often characterized as a two - step procedure : _ local _ and _ global _ assembly .
local assembly is the subject of this article .
it consists of computing the contributions of a single element in the discretized domain to the equation s approximated solution . during global assembly , these local contributions are coupled by suitably inserting them into @xmath10 and @xmath11 .
we illustrate local assembly in a concrete example , the evaluation of the local element matrix for a laplacian operator .
consider the weighted poisson equation : @xmath13 in which @xmath8 is unknown , while @xmath14 is prescribed .
the bilinear form associated with the weak variational form of the equation is : @xmath15 the domain @xmath16 of the equation is partitioned into a set of cells ( elements ) @xmath17 such that @xmath18 and @xmath19 . by defining @xmath20 as the set of basis functions with support on the element @xmath21 ( i.e. those which do not vanish on this element ) , we can express the local element matrix as @xmath22 the local element vector @xmath3 can be determined in an analogous way .
it has been shown ( for example in ) that local element tensors can be expressed as a sum of integrals over @xmath21 , each integral being the product of derivatives of functions from sets of discrete spaces and , possibly , functions of some spatially varying coefficients .
an integral of this form is called _
monomial_. quadrature schemes
are typically used to numerically evaluate @xmath23 . for convenience ,
a reference element @xmath24 and an affine mapping @xmath25 to any element @xmath26 are introduced .
this implies that a change of variables from reference coordinates @xmath27 to real coordinates @xmath28 is necessary any time a new element is evaluated .
the basis functions @xmath29 are then replaced with local basis functions @xmath30 such that @xmath31 .
the numerical integration of over an element @xmath21 can then be expressed as follows : @xmath32 where @xmath33 is the number of integration points , @xmath34 the quadrature weight at the integration point @xmath35 , @xmath36 the dimension of @xmath16 , @xmath37 the number of degrees of freedom associated to the local basis functions , and @xmath38 the determinant of the jacobian of the aforementioned change of coordinates . by exploiting the linearity , associativity and distributivity of the relevant mathematical operators , we can rewrite as @xmath39 a generalization of this transformation was introduced in @xcite . since it only involves reference element terms ,
the quadrature sum can be pre - evaluated and reused for each element .
the evaluation of the local tensor can then be abstracted as @xmath40 in which the pre - evaluated _ reference tensor _ , @xmath41 , and the cell - dependent _ geometry tensor _ , @xmath42 ,
are exposed . depending on form and discretization
, the relative performance of the two modes , in terms of the operation count , can vary quite dramatically .
the presence of derivatives or coefficient functions in the input form increases the rank of the geometry tensor , making the traditional quadrature mode preferable for sufficiently complex forms . on the other hand , speed - ups from adopting tensor mode
can be significant in a wide class of forms in which the geometry tensor remains sufficiently small .
the discretization , particularly the polynomial order of trial , test , and coefficient functions , also plays a key role in the resulting operation count .
these two modes are implemented in the fenics form compiler @xcite . in this compiler ,
a heuristic is used to choose the most suitable mode for a given form .
it consists of analysing each monomial in the form , counting the number of derivatives and coefficient functions , and checking if this number is greater than a constant found empirically @xcite .
we will return to the efficacy of this approach in section [ sec : perf - results ] .
one of the objectives of this paper is to produce a system that goes beyond the dichotomy between quadrature and tensor modes .
we will reason in terms of loop nests , code motion , and code pre - evaluation , searching the entire implementation space for an optimal synthesis .
in this section , we characterize global and local optimality for finite element integration as well as the space of legal transformations that needs be explored to achieve them .
the method by which exploration is performed is discussed in section [ sec : optimal - synthesis ] . in order to make the article
self - contained , we start with reviewing basic compiler terminology .
a perfect loop nest is a loop whose body either 1 ) comprises only a sequence of non - loop statements or 2 ) is itself a perfect loop nest .
if this condition does not hold , a loop nest is said to be imperfect .
an independent basic block is a sequence of statements such that no data dependencies exist between statements in the block .
we focus on perfect nests whose innermost loop body is an independent basic block .
a straightforward property of this class is that hoisting invariant expressions from the innermost to any of the outer loops or the preheader ( i.e. , the block that precedes the entry point of the nest ) is always safe , as long as any dependencies on loop indices are honored .
we will make use of this property .
the results of this section could also be generalized to larger classes of loop nests , in which basic block independence does not hold , although this would require refinements beyond the scope of this paper . by mapping mathematical properties to the loop nest level , we introduce the concepts of a _ linear loop _ and , more generally , a ( perfect ) multilinear loop nest .
[ def : linear - loop ] a loop @xmath3 defining the iteration space @xmath43 through the iteration variable @xmath44 , or simply @xmath45 , is linear if in its body 1 .
@xmath44 appears only as an array index , and 2 .
whenever an array @xmath2 is indexed by @xmath44 ( @xmath46 $ ] ) , all expressions in which this appears are affine in @xmath46 $ ] .
[ def : multi - linear - loop ] a multilinear loop nest of arity @xmath37 is a perfect nest composed of @xmath37 loops , in which all of the expressions appearing in the body of the innermost loop are affine in each loop @xmath45 separately .
we will show that multilinear loop nests , which arise naturally when translating bilinear or linear forms into code , are important because they have a structure that we can take advantage of to reach a local optimum .
we define two other classes of loops .
[ def : i - loop ] a loop @xmath45 is said to be a reduction loop if in its body 1 .
@xmath44 appears only as an array index , and 2 . for each augmented assignment statement @xmath47 ( e.g. , an increment )
, arrays indexed by @xmath44 appear only on the right hand side of @xmath47 .
[ def : e - loop ] a loop @xmath45 is said to be an order - free loop if its iterations can be executed in any arbitrary order .
consider equation [ eq : quadrature ] and the ( abstract ) loop nest implementing it illustrated in figure [ code : loopnest ] . the imperfect nest @xmath48 $ ] comprises an order - free loop @xmath49 ( over elements in the mesh ) , a reduction loop @xmath45 ( performing numerical integration ) , and a multilinear loop nest @xmath50 $ ] ( over test and trial functions ) . in the body of @xmath51 , one or more statements evaluate the local tensor for the element @xmath52 .
expressions ( the right hand side of a statement ) result from the translation of a form in high level matrix notation into code .
in particular , @xmath53 is the number of monomials ( a form is a sum of monomials ) , @xmath54 ( @xmath55 ) represents the product of a coefficient function ( e.g. , the inverse jacobian matrix for the change of coordinates ) with test or trial functions , and @xmath56 is a function of coefficients and geometry .
we do not pose any restrictions on function spaces ( e.g. , scalar- or vector - valued ) , coefficient expressions ( linear or non - linear ) , differential and vector operators , so @xmath56 can be arbitrarily complex .
we say that such an expression is in _ normal form _ , because the algebraic structure of a variational form is intact : products have not yet been expanded , distinct monomials can still be identified , and so on .
this brings us to formalize the class of loop nests that we aim to optimize .
[ def : fem - loopnest ] a finite element integration loop nest is a loop nest in which the following appear , in order : an imperfect order - free loop , an imperfect ( perfect only in some special cases ) , linear or non - linear reduction loop , and a multilinear loop nest whose body is an independent basic block in which expressions are in normal form .
we then characterize optimality for a finite element integration loop nest as follows .
[ def : mln - optimality ] let @xmath57 be a generic loop nest , and let @xmath58 be a transformation function @xmath59 such that @xmath60 is semantically equivalent to @xmath57 ( possibly , @xmath61 ) .
we say that @xmath62 is an optimal synthesis of @xmath57 if the total number of operations ( additions , products ) performed to evaluate the result is minimal . the concept of local optimality , which relies on the particular class of _ flop - decreasing _ transformations , is also introduced . a transformation which reduces the operation count is called flop - decreasing .
[ def : mln - quasi - optimality ] given @xmath57 , @xmath60 and @xmath58 as in definition [ def : mln - optimality ] , we say that @xmath62 is a locally optimal synthesis of @xmath57 if : * the number of operations ( additions , products ) in the innermost loops performed to evaluate the result is minimal , and * @xmath58 is expressed as composition of flop - decreasing transformations .
the restriction to flop - decreasing transformations aims to exclude those apparent optimizations that , to achieve flop - optimal innermost loops , would rearrange the computation at the level of the outer loops causing , in fact , a global increase in operation count .
we also observe that definitions [ def : mln - optimality ] and [ def : mln - quasi - optimality ] do not take into account memory requirements .
if the execution of loop nest were memory - bound the ratio of operations to bytes transferred from memory to the cpu being too low then optimizing the number of flops would be fruitless .
henceforth we assume we operate in a cpu - bound regime , evaluating arithmetic - intensive expressions . in the context of finite elements , this is often true for more complex multilinear forms and/or higher order elements .
achieving optimality in polynomial time is not generally feasible , since the @xmath56 sub - expressions can be arbitrarily unstructured .
however , multilinearity results in a certain degree of regularity in @xmath54 and @xmath55 . in the following sections
, we will elaborate on these observations and formulate an approach that achieves : ( i ) at least a local optimum in all cases ; ( ii ) global optimality whenever the monomials are `` sufficiently structured '' . to this purpose
, we will construct : * the space of legal transformations impacting the operation count ( sections [ sec : sharing - elimination ] [ sec : mem - const ] ) * an algorithm to select points in the transformation space ( section [ sec : optimal - synthesis ] ) we start with introducing the fundamental notion of sharing .
a statement within a loop nest @xmath57 presents sharing if at least one of the following conditions hold : spatial sharing : : there are at least two symbolically identical sub - expressions temporal sharing : : there is at least one non - trivial sub - expression ( e.g. , an addition or a product ) that is redundantly executed because it is independent of @xmath63 . to illustrate the definition , we show in figure [ code : multi_loopnest ] how sharing evolves as factorization and code motion
are applied to a trivial multilinear loop nest . in the original loop nest ( figure [ code : multi_loopnest_a ] ) , spatial sharing is induced by the symbol @xmath64 .
factorization eliminates spatial sharing and creates temporal sharing ( figure [ code : multi_loopnest_b ] ) . finally , generalized code motion @xcite , which hoists sub - expressions that are redundantly executed by at least one loop in the nest , leads to optimality ( figure [ code : multi_loopnest_c ] ) .
= 13pt = 13pt = 4pt in this section , we study _ sharing elimination _ , a transformation that aims to reduce the operation count by removing sharing through the application of expansion , factorization , and generalized code motion .
if the objective were reaching optimality and the expressions lacked structure , a transformation of this sort would require solving a large combinatorial problem for instance to evaluate the impact of all possible factorizations .
our sharing elimination strategy , instead , exploits the structure inherent in finite element integration expressions to guarantee , after coordination with other transformations ( an aspect which we discuss in the following sections ) , local optimality .
global optimality is achieved if stronger preconditions hold . setting local optimality , rather than optimality , as primary goal is essential to produce simple and computationally efficient algorithms two necessary conditions for integration with a compiler .
finite element expressions can be seen as composition of operations between tensors . often , the optimal implementation strategy for these operations is to be determined out of two alternatives .
for instance , consider @xmath65 , with @xmath66 being the transposed inverse jacobian matrix for the change of ( two - dimensional ) coordinates , and @xmath67 a generic two - dimensional vector .
the tensor operation will reduce to the scalar expression @xmath68 , in which @xmath69 and @xmath70 represent components of @xmath67 that depend on @xmath45 . to minimize the operation count for expressions of this kind
, we have two options : [ strategy : i ] eliminating temporal sharing through generalized code motion . [
strategy : ii ] eliminating spatial sharing first through product expansion and factorization and temporal sharing afterwards , again through generalized code motion . in the current example , we observe that , depending on the size of @xmath45 , applying strategy [ strategy : ii ] could reduce the operation count since the expression would be recast as @xmath71 and some hoistable sub - expressions would be exposed . on the other hand ,
strategy [ strategy : i ] would have no effect as @xmath67 only depends on a single loop , @xmath45 .
in general , the choice between the two strategies depends on multiple factors : the loop sizes , the increase in operation count due to expansion ( in strategy [ strategy : ii ] ) , and the gain due to code motion .
a second application of strategy [ strategy : ii ] was provided in figure [ code : multi_loopnest ] .
these examples motivate the introduction of a particular class of expressions , for which the two strategies assume notable importance .
[ def : struct - expr ] we say that an expression is `` structured along a loop nest @xmath57 '' if and only if , for every symbol @xmath72 depending on at least one loop in @xmath57 , the spatial sharing of @xmath72 may be eliminated by factorizing all occurrences of @xmath72 in the expression . [ prop : multi - struct ] an expression along a multilinear loop nest is structured .
this follows directly from definition [ def : linear - loop ] and definition [ def : multi - linear - loop ] , which essentially restrict the number of occurrences of a symbol @xmath72 in a summand to at most 1 .
if @xmath57 were an arbitrary loop nest , a given symbol @xmath72 could appear everywhere ( e.g. , @xmath37 times in a summand and @xmath53 times in another summand with @xmath73 , as argument of a higher level function , in the denominator of a division ) , thus posing the challenge of finding the factorization that maximizes temporal sharing . if @xmath57 is instead a finite element integration loop nest , thanks to proposition [ prop : multi - struct ] the space of flop - decreasing transformations is constructed by `` composition '' of strategy [ strategy : i ] and strategy [ strategy : ii ] , as illustrated in algorithm [ algo : sharing - elimination ] .
finally , we observe that the @xmath56 sub - expressions can sometimes be considered `` weakly structured '' .
this happens when a relaxed version of definition [ def : struct - expr ] applies , in which the factorization of @xmath72 only `` minimizes '' ( rather than `` eliminates '' ) spatial sharing ( for instance , in the complex hyperelastic model analyzed in section [ sec : perf - results ] )
. weak structure will be exploited by algorithm [ algo : sharing - elimination ] in the attempt to achieve optimality .
algorithm [ algo : sharing - elimination ] describes sharing elimination assuming as input a tree representation of the loop nest .
it makes use of the following notation and terminology : * _ multilinear operand _ : any @xmath54 or @xmath55 in the input expression . *
_ multilinear symbol _ : a symbol appearing within a multilinear operand depending on @xmath74 or @xmath51 ( e.g. , test functions , first order derivatives of test functions , etc . ) .
examples will be provided in section [ sec : se - examples ] . ' '' '' [ algo : sharing - elimination ] the input of the algorithm is a tree representation a finite element integration loop nest . 1 . perform a depth - first visit of the loop tree to collect and partition multilinear operands into disjoint sets , @xmath75 .
@xmath76 is such that all multilinear operands in each @xmath77 share the same set of multilinear symbols @xmath78 , whereas there is no sharing across different partitions . for all multilinear operands in @xmath77 such that @xmath79 , apply strategy [ strategy : i ] .
+ _ note : as a consequence of proposition [ prop : multi - struct ] , @xmath80 and @xmath81 represent the number of products in the innermost loop induced by @xmath82 if strategy [ strategy : i ] or strategy [ strategy : ii ] were applied _ 2 . for each sub - expression @xmath83 depending on exactly one linear loop ,
collect the multilinear symbols and the temporaries produced at step ( 1 ) .
partition them into disjoint sets , @xmath84 , such that @xmath85 includes all instances of a given symbol in @xmath83 .
apply strategy [ strategy : ii ] factorizing the symbols in each @xmath85 , provided that this leads to a reduction in operation count ; otherwise , apply strategy [ strategy : i ] + _ note : the last check ensures the flop - decreasing nature of the transformation . in the cases in which expansion outweighs code motion , strategy [ strategy : i ] is preferred . _
+ _ note : the expansion cost is a function of the products wrapping a symbol ( how many of them and their arity ) , so it can be determined through tree visits .
build the _ sharing graph _ @xmath86 .
each @xmath87 represents a multilinear symbol or a temporary produced by the previous steps .
an edge @xmath88 , @xmath89 indicates that a product @xmath90 would appear if the sub - expressions including @xmath91 and @xmath92 were expanded .
+ _ note : the following steps will only impact bilinear forms , since otherwise @xmath93 . _ 4 .
partition @xmath47 into disjoint sets , @xmath94 , such that @xmath95 includes all instances of a given symbol @xmath96 in the expression . transform @xmath97 by merging @xmath98 into a unique vertex @xmath96 ( taking the union of the edges ) , provided that factorizing @xmath99 $ ] would not cause an increase in operation count .
map @xmath97 to an integer linear programming ( ilp ) model for determining how to optimally apply strategy [ strategy : ii ] .
the solution is the set of symbols that will be factorized by strategy [ strategy : ii ] .
let @xmath100 ; the ilp model then is as follows : @xmath101 6 .
perform a depth - first visit of the loop tree and , for each yet unhandled or hoisted expression , apply the most profitable between strategy [ strategy : i ] and strategy [ strategy : ii ] .
+ _ note : this pass speculatively assumes that expressions are ( weakly ) structured along the reduction loop . if the assumption does not hold , the operation count will generally be sub - optimal because only a subset of factorizations and code motion opportunities may eventually be considered . _ ' '' '' although the primary goal of algorithm [ algo : sharing - elimination ] is operation count minimization within the multilinear loop nest , the enforcement of flop - decreasing transformations ( steps ( 2 ) and ( 4 ) ) and the re - scheduling of sub - expressions within outer loops ( last step ) also attempt to optimize the loop nest globally .
we will further elaborate this aspect in section [ sec : proof ] .
consider again figure [ code : multi_loopnest_a ] .
we have @xmath102 , with @xmath103 , @xmath104 , and @xmath105 . for all @xmath106
, we have @xmath107 , although applying strategy [ strategy : i ] in step ( 1 ) has no effect .
the sharing graph is @xmath108 , and @xmath109 .
the ilp formulation leads to the code in figure [ code : multi_loopnest_c ] . in figure
[ code : poisson ] , algorithm [ algo : sharing - elimination ] is executed in a very simple realistic scenario , which originates from the bilinear form of a poisson equation in two dimensions .
we observe that @xmath110 , with @xmath111 and @xmath112 .
in addition , @xmath113 , so strategy [ strategy : i ] is applied to both partitions ( step ( 1 ) ) .
we then have ( step ( 3 ) ) @xmath114 .
since there are no more factorization opportunities , the ilp formulation becomes irrelevant .
= 7pt = 7pt for reasons of space , further examples , including the hyperelastic model evaluated in section [ sec : perf - results ] and other non - trivial ilp instances , are made available online .
sharing elimination uses three operators : expansion , factorization , and code motion . in this section ,
we discuss the role and legality of a fourth operator : reduction pre - evaluation .
we will see that what makes this operator special is the fact that there exists a single point in the transformation space of a monomial ( i.e. , a specific factorization of test , trial , and coefficient functions ) ensuring its correctness .
we start with an example .
consider again the loop nest and the expression in figure [ code : loopnest ] .
we pose the following question : are we able to identify sub - expressions for which the reduction induced by @xmath45 can be pre - evaluated , thus obtaining a decrease in operation count proportional to the size of @xmath45 , @xmath43 ?
the transformation we look for is exemplified in figure [ code : loopnest_rednored ] with a simple loop nest .
the reader may verify that a similar transformation is applicable to the example in figure [ code : poisson_a ] .
= 19pt = 5pt pre - evaluation can be seen as the generalization of tensor contraction ( section [ sec : tc ] ) to a wider class of sub - expressions .
we know that multilinear forms can be seen as sums of monomials , each monomial being an integral over the equation domain of products ( of derivatives ) of functions from discrete spaces . a monomial can always be reduced to the product between a `` reference '' and a `` geometry '' tensor . in our model , a reference tensor is simply represented by one or more sub - expressions independent of @xmath49 , exposed after particular transformations of the expression tree .
this leads to the following algorithm . ' '' ''
[ algo : pre - evaluation ] consider a finite element integration loop nest @xmath115 $ ] .
we dissect the normal form input expression into distinct sub - expressions , each of them representing a monomial .
each sub - expression is then factorized so as to split constants from @xmath116$]-dependent terms .
this transformation is feasible , as a consequence of the results in .
these @xmath116$]-dependent terms are hoisted outside of @xmath57 and stored into temporaries . as part of this process ,
the reduction induced by @xmath45 is computed by means of symbolic execution .
finally , @xmath45 is removed from @xmath57 . ' '' '' the pre - evaluation of a monomial introduces some critical issues : 1 . depending on the complexity of a monomial
, a certain number , @xmath117 , of temporary variables is required if pre - evaluation is performed .
such temporary variables are actually @xmath37-dimensional arrays of size @xmath47 , with @xmath37 and @xmath47 being , respectively , the arity and the extent ( iteration space size ) of the multilinear loop nest ( e.g. , @xmath118 and @xmath119 in the case of bilinear forms ) . for certain values of @xmath120 , pre - evaluation may dramatically increase the working set , which may be counter - productive for actual execution time
the transformations exposing @xmath116$]-dependent terms increase the arithmetic complexity of the expression ( e.g. , expansion tends to increase the operation count ) .
this could outweigh the gain due to pre - evaluation .
3 . a strategy for coordinating sharing elimination and pre - evaluation is needed .
we observe that sharing elimination inhibits pre - evaluation , whereas pre - evaluation could expose further sharing elimination opportunities .
we expand on point ( 1 ) in the next section , while we address points ( 2 ) and ( 3 ) in section [ sec : optimal - synthesis ] .
we have just observed that the code motion induced by monomial pre - evaluation may dramatically increase the working set size .
even more aggressive code motion strategies are theoretically conceivable .
imagine @xmath57 is enclosed in a time stepping loop .
one could think of exposing ( through some transformations ) and hoisting time - invariant sub - expressions for minimizing redundant computation at each time step .
the working set size would then increase by a factor @xmath121 , and since @xmath122 , the gain in operation count would probably be outweighed , from a runtime viewpoint , by a much larger memory pressure . since , for certain forms and discretizations , hoisting may cause the working set to exceed the size of some level of local memory ( e.g. the last level of private cache on a conventional cpu , the shared memory on a gpu )
, we introduce the following _ memory constraints_. [ const : le ] the size of a temporary due to code motion must not be proportional to the size of @xmath49 .
[ const : th ] the total amount of memory occupied by the temporaries due to code motion must not exceed a certain threshold , ` t_h ` . constraint [ const : le ] is a policy decision that the compiler should not silently consume memory on global data objects .
it has the effect of shrinking the transformation space .
constraint [ const : th ] has both theoretical and practical implications , which will be carefully analyzed in the next sections .
in this section , we build a transformation algorithm that , given a memory bound , systematically reaches a local optimum for finite element integration loop nests .
we address the two following issues : 1 .
_ coordination of pre - evaluation and sharing elimination .
_ recall from section [ sec : pre - evaluation ] that pre - evaluation could either increase or decrease the operation count in comparison with that achieved by sharing elimination .
2 . _ optimizing over composite operations . _ consider a form comprising two monomials @xmath123 and @xmath124 .
assume that pre - evaluation is profitable for @xmath123 but not for @xmath124 , and that @xmath123 and @xmath124 share at least one term ( for example some basis functions ) . if pre - evaluation were applied to @xmath123 , sharing between @xmath123 and @xmath124 would be lost .
we then need a mechanism to understand which transformation
pre - evaluation or sharing elimination results in the highest operation count reduction when considering the whole set of monomials ( i.e. , the expression as a whole ) .
let @xmath125 be a cost function that , given a monomial @xmath126 , returns the gain / loss achieved by pre - evaluation over sharing elimination .
in particular , we define @xmath127 , where @xmath128 and @xmath129 represent the operation counts resulting from applying sharing elimination and pre - evaluation , respectively . thus pre - evaluation is profitable for @xmath53 if and only if @xmath130 .
we return to the issue of deriving @xmath128 and @xmath129 in section [ sec : op_count ] . having defined @xmath131
, we can now describe the transformation algorithm ( algorithm [ algo : gamma ] ) . ' '' '' [ algo : gamma ] the algorithm has three main phases : initialization ( step 1 ) ; determination of the monomials preserving the memory constraints that should be pre - evaluated ( steps 2 - 4 ) ; application of pre - evaluation and sharing elimination ( step 5 ) . 1 . perform a depth - first visit of the expression tree and determine the set of monomials @xmath132 .
let @xmath47 be the subset of monomials @xmath53 such that @xmath133 .
the set of monomials that will _ potentially _ be pre - evaluated is @xmath134 .
+ _ note : there are two fundamental reasons for not pre - evaluating @xmath135 straight away : 1 ) the potential presence of spatial sharing between @xmath123 and @xmath136 , which impacts the search for the global optimum ; 2 ) the risk of breaking constraint [ const : th ] .
build the set @xmath137 of all possible bipartitions of @xmath82 .
let @xmath138 be the dictionary that will store the operation counts of different alternatives .
3 . discard @xmath139 if the memory required after applying pre - evaluation to the monomials in @xmath140 exceeds @xmath141 ( see constraint [ const : th ] ) ; otherwise , add @xmath142 = \mathrm{\theta}^{se}(s \cup b_s ) + \mathrm{\theta}^{pre}(b_p)$ ] .
+ _ note : @xmath143 is in practice very small , since even complex forms usually have only a few monomials .
this pass can then be accomplished rapidly as long as the cost of calculating @xmath128 and @xmath129 is negligible .
we elaborate on this aspect in section [ sec : op_count ] . _
take @xmath144 $ ] .
5 . apply pre - evaluation to all monomials in @xmath140 .
apply sharing elimination to all resulting expressions .
+ _ note : because of the reuse of basis functions , pre - evaluation may produce some identical tables , which will be mapped to the same temporary variable .
sharing elimination is therefore transparently applied to all expressions , including those resulting from pre - evaluation . _ ' '' '' the output of the transformation algorithm is provided in figure [ code : loopnest - opt ] , assuming as input the loop nest in figure [ code : loopnest ] .
we tie up the remaining loose end : the construction of the cost function @xmath131 .
we recall that @xmath145 , with @xmath128 and @xmath129 representing the operation counts after applying sharing elimination and pre - evaluation .
since @xmath131 is deployed in a working compiler , simplicity and efficiency are essential characteristics . in the following ,
we explain how to derive these two values .
the most trivial way of evaluating @xmath128 and @xmath129 would consist of applying the actual transformations and simply count the number of operations .
this would be tolerable for @xmath128 , as algorithm [ algo : sharing - elimination ] tends to have negligible cost .
however , the overhead would be unacceptable if we applied pre - evaluation in particular , symbolic execution to all bipartitions analyzed by algorithm [ algo : gamma ] .
we therefore seek an analytic way of determining @xmath129 .
the first step consists of estimating the _ increase factor _ , @xmath146 .
this number captures the increase in arithmetic complexity due to the transformations exposing pre - evaluation opportunities . for context , consider the example in figure [ code : increase_factor ] .
one can think of this as the ( simplified ) loop nest originating from the integration of the action of a mass matrix .
the sub - expression ` f_0*b_{i0}+f_1*b_{i1}+f_2*b_{i2 } ` represents the coefficient @xmath147 over ( tabulated ) basis functions ( array @xmath137 ) . in order to apply pre - evaluation
, the expression needs be transformed to separate @xmath147 from all @xmath116$]-dependent quantities ( see algorithm [ algo : pre - evaluation ] ) . by product expansion , we observe an increase in the number of @xmath50$]-dependent terms of a factor @xmath148 . in general , however , determining @xmath146 is not so straightforward since redundant tabulations may result from common sub - expressions . consider the previous example .
one may add one coefficient in the same function space as @xmath147 , repeat the expansion , and observe that multiple sub - expressions ( e.g. , @xmath149 and @xmath150 ) will reduce to identical tables .
to evaluate @xmath146 , we then use combinatorics .
we calculate the @xmath151-combinations with repetitions of @xmath37 elements , where : ( i ) @xmath151 is the number of ( derivatives of ) coefficients appearing in a product ; ( ii ) @xmath37 is the number of unique basis functions involved in the expansion . in the original example
, we had @xmath152 ( for @xmath153 , @xmath154 , and @xmath155 ) and @xmath156 , which confirms @xmath157 . in the modified example
, there are two coefficients , so @xmath158 , which means @xmath159 .
if @xmath160 ( the extent of the reduction loop ) , we already know that pre - evaluation will not be profitable . intuitively , this means that we are introducing more operations than we are saving from pre - evaluating @xmath45 .
if @xmath161 , we still need to find the number of terms @xmath162 such that @xmath163 .
the mass matrix monomial in figure [ code : increase_factor ] is characterized by the dot product of test and trial functions , so trivially @xmath164 . in the example in figure [ code : poisson ] , instead , we have @xmath165 after a suitable factorization of basis functions . in general , therefore , @xmath162 depends on both form and discretization .
to determine this parameter , we look at the re - factorized expression ( as established by algorithm [ algo : pre - evaluation ] ) , and simply count the terms amenable to pre - evaluation .
we demonstrate that the orchestration of sharing elimination and pre - evaluation performed by the transformation algorithm guarantees local optimality ( definition [ def : mln - quasi - optimality ] ) .
the proof re - uses concepts and explanations provided throughout the paper , as well as the terminology introduced in section [ sec : se - algo ] .
[ prop : optimal - approach ] consider a multilinear form comprising a set of monomials @xmath132 , and let @xmath57 be the corresponding finite element integration loop nest .
let @xmath58 be the transformation algorithm .
let @xmath27 be the set of monomials that , according to @xmath58 , need to be pre - evaluated , and let @xmath166 .
assume that the pre - evaluation of different monomials does not result in identical tables .
then , @xmath167 is a local optimum in the sense of definition [ def : mln - quasi - optimality ] and satisfies constraint [ const : th ] .
we first observe that the cost function @xmath131 predicts the _ exact _ gain / loss in monomial pre - evaluation , so @xmath27 and @xmath168 can actually be constructed .
let @xmath169 denote the operation count for @xmath57 and let @xmath170 be the subset of innermost loops ( all @xmath51 loops in figure [ code : loopnest - opt ] ) .
we need to show that there is no other synthesis @xmath171 satisfying constraint [ const : th ] such that @xmath172 .
this holds if and only if 1 .
_ the coordination of pre - evaluation with sharing elimination is optimal_. this boils down to prove that 1 .
_ pre - evaluating any @xmath173 would result in @xmath174 _ 2 .
_ not pre - evaluating any @xmath175 would result in @xmath174 _ 2 .
_ sharing elimination leads to a ( at least ) local optimum . _ + we discuss these points separately 1 . 1 .
let @xmath176 represent the set of tables resulting from applying pre - evaluation to a monomial @xmath53 .
consider two monomials @xmath177 and the respective sets of pre - evaluated tables , @xmath178 and @xmath179 .
if @xmath180 , at least one table is assignable to the same temporary . @xmath58 , therefore , may not be optimal , since @xmath131 only distinguishes monomials in `` isolation '' .
we neglect this scenario ( see assumptions ) because of its purely pathological nature and its with high probability
negligible impact on the operation count .
2 . let @xmath181 and @xmath182 be two monomials sharing some generic multilinear symbols . if @xmath123 were carelessly pre - evaluated , there may be a potential gain in sharing elimination that is lost , potentially leading to a non - optimum .
this situation is prevented by construction , because @xmath58 exhaustively searches all possible bipartitions on order to determine an optimum which satisfies constraint [ const : th ] .
recall that since the number of monomials is in practice very small , this pass can rapidly be accomplished .
2 . consider algorithm [ algo : sharing - elimination ] .
proposition [ prop : multi - struct ] ensures that there are only two ways of scheduling the multilinear operands in @xmath77 : through generalized code motion ( strategy [ strategy : i ] ) or factorization of multilinear symbols ( via strategy [ strategy : ii ] ) . if applied , these two strategies would lead , respectively , to performing @xmath80 and @xmath81 multiplications at every loop iteration .
since strategy [ strategy : i ] is applied if and only if @xmath183 and does not change the structure of the expression ( it requires neither expansion nor factorization ) , step ( 1 ) can not prune the optimum from the search space .
+ after structuring the sharing graph @xmath97 in such a way that only flop - decreasing transformations are possible , the ilp model is instantiated . at this point , proving optimality reduces to establishing the correctness of the model , which is relatively straightforward because of its simplicity .
the model aims to minimize the operation count by selecting the most promising factorizations .
the second set of constraints is to select all edges ( i.e. , all multiplications ) , exactly once .
the first set of inequalities allows multiplications to be scheduled : once a vertex @xmath96 is selected ( i.e. , once a symbol is decided to be factorized ) , all multiplications involving @xmath96 can be grouped . throughout the paper
we have reiterated the claim that algorithm [ algo : gamma ] achieves a globally optimal flop count if stronger preconditions on the input variational form are satisfied .
we state here these preconditions , in increasing order of complexity . 1 .
there is a single monomial and only a specific coefficient ( e.g. , the coordinates field ) .
this is by far the simplest scenario , which requires no particular transformation at the level of the outer loops , so optimality naturally follows .
2 . there is a single monomial , but multiple coefficients are present .
optimality is achieved if and only if all sub - expressions depending on coefficients are structured ( see section [ sec : se - rln ] ) .
this avoids ambiguity in factorization , which in turn guarantees that the output of step ( 7 ) in algorithm [ algo : sharing - elimination ] is optimal .
3 . there are multiple monomials , but either at most one coefficient ( e.g. , the coordinates field ) or multiple coefficients not inducing sharing across different monomials are present .
this reduces , respectively , to cases ( 1 ) and ( 2 ) above .
there are multiple monomials , and coefficients are shared across monomials .
optimality is reached if and only if the coefficient - dependent sub - expressions produced by algorithm [ algo : sharing - elimination ] that is , the by - product of factorizing test / trial functions from distinct monomials preserve structure .
sharing elimination and pre - evaluation , as well as the transformation algorithm , have been implemented in coffee , the compiler for finite element integration routines adopted in firedrake . in this section
, we briefly discuss the aspects of the compiler that are relevant for this article .
coffee implements sharing elimination and pre - evaluation by composing building block transformation operators , which we refer to as _ rewrite operators_. this has several advantages .
the first is extensibility .
new transformations , such as sum factorization in spectral methods , could be expressed by composing the existing operators , or with small effort building on what is already available .
second , generality : coffee can be seen as a lightweight , low level computer algebra system , not necessarily tied to finite element integration .
third , robustness : the same operators are exploited , and therefore tested , by different optimization pipelines .
the rewrite operators , whose ( python ) implementation is based on manipulation of abstract syntax trees ( asts ) , comprise the coffee language .
a non - exhaustive list of such operators includes expansion , factorization , re - association , generalized code motion .
coffee aims to be independent of the high level form compiler .
it provides an interface to build generic asts and only expects expressions to be in normal form ( or sufficiently close to it ) .
for example , firedrake has transitioned from a version of the fenics form compiler @xcite modified to produce asts rather than strings , to a newly written compiler , while continuing to emply coffee .
thus , coffee decouples the mathematical manipulation of a form from code optimization ; or , in other words , relieves form compiler developers of the task of fine scale loop optimization of generated code . for several reasons
, basis function tables may be block - sparse ( e.g. , containing zero - valued columns ) .
for example , the fenics form compiler implements vector - valued functions by adding blocks of zero - valued columns to the corresponding tabulations ; this extremely simplifies code generation ( particularly , the construction of loop nests ) , but also affects the performance of the generated code due to the execution of `` useless '' flops ( e.g. , operations like ` a + 0 ` ) . in , a technique to avoid iteration over zero - valued columns based on the use of indirection arrays ( e.g. ` a[b[i ] ] ` , in which ` a ` is a tabulated basis function and ` b ` a map from loop iterations to non - zero columns in a ) was proposed .
this technique , however , produces non - contiguous memory loads and stores , which nullify the potential benefits of vectorization .
coffee , instead , handles block - sparse basis function tables by restructuring loops in such a manner that low level optimization ( especially vectorization ) is only marginally affected .
this is based on symbolic execution of the code , which enables a series of checks on array indices and loop bounds which determine the zero - valued blocks which can be skipped without affecting data alignment .
experiments were run on a single core of an intel i7 - 2600 ( sandy bridge ) cpu , running at 3.4ghz , 32 kb l1 cache ( private ) , 256 kb l2 cache ( private ) and 8 mb l3 cache ( shared ) .
the intel turbo boost and intel speed step technologies were disabled .
the intel ` icc 15.2 ` compiler was used .
the compilation flags used were ` -o3 , -xhost ` .
the compilation flag ` xhost ` tells the intel compiler to generate efficient code for the underlying platform .
the zenodo system was used to archive all packages used to perform the experiments : firedrake @xcite , petsc @xcite , petsc4py @xcite , fiat @xcite , ufl @xcite , ffc @xcite , pyop2 @xcite and coffee @xcite .
the experiments can be reproduced using a publicly available benchmark suite @xcite .
we analyze the execution time of four real - world bilinear forms of increasing complexity , which comprise the differential operators that are most common in finite element methods .
in particular , we study the mass matrix ( `` ` mass ` '' ) and the bilinear forms arising in a helmholtz equation ( `` ` helmholtz ` '' ) , in an elastic model ( `` ` elasticity ` '' ) , and in a hyperelastic model ( `` ` hyperelasticity ` '' ) .
the complete specification of these forms is made publicly available .
we evaluate the speed - ups achieved by a wide variety of transformation systems over the `` original '' code produced by the fenics form compiler ( i.e. , no optimizations applied ) .
we analyze the following transformation systems : quad : : optimized quadrature mode .
work presented in , implemented in in the fenics form compiler .
tens : : tensor contraction mode .
work presented in , implemented in the fenics form compiler .
auto : : automatic choice between ` tens ` and ` quad ` driven by heuristic ( detailed in and summarized in section [ sec : qualitative ] ) . implemented in the fenics form compiler .
ufls : : uflacs , a novel back - end for the fenics form compiler whose primary goals are improved code generation and execution times .
cfo1 : : generalized loop - invariant code motion .
work presented in , implemented in coffee .
cfo2 : : optimal loop nest synthesis with handling of block - sparse tables . work presented in this article , implemented in coffee .
the values that we report are the average of three runs with `` warm cache '' ; that is , with all kernels retrieved directly from the firedrake s cache , so code generation and compilation times are not counted .
the timing includes however the cost of both local assembly and matrix insertion , with the latter minimized through the choice of a mesh ( details below ) small enough to fit the l3 cache of the cpu . for a fair comparison ,
small patches were written to the make ` quad ` , ` tens ` , and ` ufls ` compatible with firedrake . by executing all simulations in firedrake , we guarantee that both matrix insertion and mesh iteration have a fixed cost , independent of the transformation system employed .
the patches adjust the data storage layout to what firedrake expects ( e.g. , by generating an array of pointers instead of a pointer to pointers , by replacing flattened arrays with bi - dimensional ones ) . for constraint [ const : th ] , discussed in section [ sec : mem - const ]
, we set @xmath184 ; that is , the size of the processor l2 cache ( the last level of private cache ) . when the threshold had an impact on the transformation process , the experiments were repeated with @xmath185 .
the results are documented later , individually for each problem . following the methodology adopted in , we vary the following parameters : * the polynomial degree of test , trial , and coefficient ( or `` pre - multiplying '' ) functions , @xmath186 * the number of coefficient functions @xmath187 while constants of our study are * the space of test , trial , and coefficient functions : lagrange * the mesh : tetrahedral with a total of 4374 elements * exact numerical quadrature ( we employ the same scheme used in , based on the gauss - legendre - jacobi rule ) we report the results of our experiments in figures [ fig : mass ] , [ fig : helmholtz ] , [ fig : elasticity ] , and [ fig : hyperelasticity ] as three - dimensional plots .
the axes represent @xmath188 , @xmath189 , and code transformation system .
we show one subplot for each problem instance @xmath190 , with the code transformation system varying within each subplot . the best variant for each problem instance is given by the tallest bar , which indicates the maximum speed - up over non - transformed code
. we note that if a bar or a subplot are missing , then the form compiler failed to generate code because it either exceeded the system memory limit or was otherwise unable to handle the form .
the rest of the section is organized as follows : we first provide insights into the general outcome of the experimentation ; we then comment on the impact of a fundamental low - level optimization , namely autovectorization ; finally , we motivate , for each form , the performance results obtained .
[ [ high - level - view ] ] high level view + + + + + + + + + + + + + + + our transformation strategy does not always guarantee minimum execution time . in particular , about 5@xmath191 of the test cases ( 3 out of 56 , without counting marginal differences )
show that ` cfo2 ` was not optimal in terms of runtime .
the most significant of such test cases is the elastic model with @xmath192 $ ] .
there are two reasons for this .
first , low level optimization can have a significant impact on the actual performance .
for example , the aggressive loop unrolling in ` tens ` eliminates operations on zeros and reduces the working set size by not storing entire temporaries ; on the other hand , preserving the loop structure can maximize the chances of autovectorization .
second , the transformation strategy adopted when @xmath141 is exceeded plays a key role , as we will later elaborate .
[ [ autovectorization ] ] autovectorization + + + + + + + + + + + + + + + + + we chose the mesh dimension and the function spaces such that the inner loop sizes would always be a multiple of the machine vector length .
this ensured autovectorization in the majority of code variants .
the biggest exception is ` quad ` , due to the presence of indirection arrays in the generated code . in ` tens ` , loop nests are fully unrolled , so the standard loop vectorization is not feasible ; the compiler reports suggest , however , that block vectorization @xcite is often triggered . in ` ufls ` , ` cfo1 ` , and ` cfo2 ` the iteration spaces have identical structure , with loop vectorization being regularly applied . [
[ mass - matrix ] ] mass matrix + + + + + + + + + + + we start with the simplest of the bilinear forms investigated , the mass matrix .
results are in figure [ fig : mass ] .
we first notice that the lack of improvements when @xmath193 is due to the fact that matrix insertion outweighs local assembly . for @xmath194 , ` cfo2 ` generally shows the highest speed - ups .
it is worth noting why ` auto ` does not always select the fastest implementation : ` auto ` always opts for ` tens ` , while for @xmath195 ` quad ` tends to be preferable . on the other hand , ` cfo2 ` always makes the optimal decision about whether to apply pre - evaluation or not . surprisingly , despite the simplicity of the form , the performance of the various code generation systems can differ significantly . [ [ helmholtz ] ] helmholtz + + + + + + + + + as in the case of mass matrix , when @xmath193 the matrix insertion phase is dominant . for @xmath194 ,
the general trend is that ` cfo2 ` outperforms the competitors . in particular : @xmath196 : : pre - evaluation makes ` cfo2 ` notably faster than ` cfo1 ` , especially for high values of @xmath188 ; ` auto ` correctly selects ` tens ` , which is comparable to ` cfo2 ` . @xmath197
: : ` auto ` picks ` tens ` ; the choice is however sub - optimal when @xmath198 and @xmath199
. this can indirectly be inferred from the large gap between ` cfo2 ` and ` tens / auto ` : ` cfo2 ` applies sharing elimination , but it correctly avoids pre - evaluation because of the excessive expansion cost . @xmath200 and
@xmath201 : : ` auto ` reverts to ` quad ` , which would theoretically be the right choice ( the flop count is much lower than in ` tens ` ) ; however , the generated code suffers from the presence of indirection arrays , which break autovectorization and `` traditional '' code motion .
the slow - downs ( or marginal improvements ) seen in a small number of cases exhibited by ` ufls ` can be attributed to the presence of sharing in the generated code .
an interesting experiment we additionally performed was relaxing the memory threshold by setting @xmath185 .
we found that this makes ` cfo2 ` generally slower for @xmath195 , with a maximum slow - down of 2.16@xmath202 with @xmath203 .
this effect could be worse when running in parallel , since the l3 cache is shared and different threads would end up competing for the same resource . [ [ elasticity ] ] elasticity + + + + + + + + + + the results for the elastic model are displayed in figure [ fig : elasticity ] .
the main observation is that ` cfo2 ` never triggers pre - evaluation , although in some occasions it should . to clarify this ,
consider the test case @xmath204 , in which ` tens / auto ` show a considerable speed - up over ` cfo2 ` . `
cfo2 ` finds pre - evaluation profitable in terms of operation count , although it is eventually not applied to avoid exceeding @xmath141 .
however , running the same experiments with @xmath185 resulted in a dramatic improvement , even higher than that obtained by ` tens ` .
the reason is that , despite exceeding @xmath141 by roughly 40@xmath191 , the saving in operation count is so large ( 5@xmath202 in this specific problem ) that pre - evaluation would in practice be the winning choice .
this suggests that our objective function should be improved to handle the cases in which there is a significant gap between potential cache misses and reduction in operation count .
we also note that : * the differences between ` cfo2 ` and ` cfo1 ` are due to the perfect sharing elimination and the zero - valued blocks avoidance technique presented in section [ sec : zeros ] .
* when @xmath197 , ` auto ` prefers ` tens ` over ` quad ` , which leads to sub - optimal operation counts and execution times . * ` ufls ` often results in better execution times than ` quad ` and ` tens ` .
this is due to multiple factors , including avoidance of indirection arrays , preservation of loop structure , and a more effective code motion strategy .
[ [ hyperelasticity ] ] hyperelasticity + + + + + + + + + + + + + + + in the experiments on the hyperelastic model , shown in figure [ fig : hyperelasticity ] , ` cfo2 ` exhibits the largest gains out of all problem instances considered in this paper .
this is a positive result , since it indicates that our transformation algorithm scales well with form complexity .
the fact that all code transformation systems ( apart from ` tens ` ) show quite significant speed - ups suggests two points .
first , the baseline is highly inefficient . with forms as complex as in the hyperelastic model , a trivial translation of integration routines into code
should always be avoided as even the best general - purpose compiler available ( the intel compiler on an intel platform at maximum optimization level ) fails to exploit the structure inherent in the expressions .
second , the strategy for removing spatial and temporal sharing has a tremendous impact .
sharing elimination as performed by ` cfo2 ` ensures a critical reduction in operation count , which becomes particularly pronounced for higher values of @xmath188 .
we have developed a theory for the optimization of finite element integration loop nests .
the article details the domain properties which are exploited by our approach ( e.g. , linearity ) and how these translate to transformations at the level of loop nests .
all of the algorithms shown in this paper have been implemented in coffee , a compiler publicly available fully integrated with the firedrake framework .
the correctness of the transformation algorithm was discussed .
the performance results achieved suggest the effectiveness of our methodology .
we have defined sharing elimination and pre - evaluation as high level transformations on top of a specific set of rewrite operators , such as code motion and factorization , and we have used them to construct the transformation space .
there are three main limitations in this process .
first , we do not have a systematic strategy to optimize sub - expressions which are independent of linear loops .
although we have a mechanism to determine how much computation should be hoisted to the level of the integration ( reduction ) loop , it is not clear how to effectively improve the heuristics used at step ( 6 ) in algorithm [ algo : sharing - elimination ] .
second , lower operation counts may be found by exploiting domain - specific properties , such as redundancies in basis functions ; this aspect is completely neglected in this article .
third , with constraint [ const : le ] we have limited the applicability of code motion .
this constraint was essential given the complexity of the problem tackled .
another issue raised by the experimentation concerns selecting a proper threshold for constraint [ const : th ] . to solve
this problem would require a more sophisticated cost model , which is an interesting question deserving further research .
we also identify two additional possible research directions : a complete classification of forms for which a global optimum is achieved ; and a generalization of the methodology to other classes of loop nests , for instance those arising in spectral element methods . | we present an algorithm for the optimization of a class of finite element integration loop nests .
this algorithm , which exploits fundamental mathematical properties of finite element operators , is proven to achieve a locally optimal operation count . in specified circumstances
the optimum achieved is global .
extensive numerical experiments demonstrate significant performance improvements over the state of the art in finite element code generation in almost all cases .
this validates the effectiveness of the algorithm presented here , and illustrates its limitations .
this work was supported by by the department of computing at imperial college london , the engineering and physical sciences research council [ grant number ep / l000407/1 ] , and the natural environment research council [ grant numbers ne / k008951/1 and ne / k006789/1 ] , and by a hipeac collaboration grant .
the authors would like to thank dr .
andrew t.t .
mcrae , dr . lawrence mitchell , and dr .
francis russell for their invaluable suggestions and their contribution to the firedrake project .
author s addresses : fabio luporini @xmath0 paul h. j. kelly , department of computing , imperial college london ; david a. ham , department of mathematics , imperial college london ; |
zno has been extensively studied , both theoretically and experimentally , due to its many promising applications in piezoelectric devices , transistors , photodiodes , photocatalysis and antibacterial function @xcite .
the physical properties of zno , especially its surface properties , can be experimentally modified at the atomic level for the purpose of , e.g. , engineering the material for desired functionality . since zno contacts with its external environment through its surfaces , monitoring
how the surface properties respond to external perturbation ( e.g. thermal treatment ) can provide valuable insights to improve our ability to manipulate zno for application purposes .
one of the simplest way to experimentally modify the surfaces of zno is by heating it to elevated temperature ( but not exceeding its melting point ) .
heating zno can be easily carried out in practice , and many works had been reported along this line @xcite .
zno crystals are dominated by four low miller indices surfaces : the non - polar ( 1 0 @xmath4 0 ) and ( 1 1 @xmath5 0 ) surfaces and the polar surfaces which are comprised of zinc - terminated surface ( 0 0 0 1 ) and the oxygen terminated surface ( 0 0 0 @xmath4 ) .
the polar surface of ionic crystals will have a surface energy that diverges with sample size due to the generation of a macroscopic electrostatic field across the crystal @xcite , or the cleavage energy is infinite in the ionic crystals .
this kind of behavior was well investigated by tasker @xcite .
this is why wurzite zno is labeled as tusker - type surfaces , which are formed by alternating layers of oppositely charged ions . in this work
we consider specifically the ( 0 0 0 1 ) surface of a zno crystal in wurtzite structure and question what will happen to the atomic configuration of the surface when a zno slab with finite thickness is heated without melting .
sublimation of atoms from the polar surfaces , driven by temperature effect , was studied using molecular dynamics ( md ) simulation , where the trajectories of all atoms at a given temperature @xmath2 were followed quantitatively . in the md simulated annealing , in which the reactive force field ( reaxff ) for zno is used ,
sublimation of o atoms from zno polar surface is observed .
the occurrence of sublimation was attributed to the novelty of zno reaxff that allows for bond formation and charge transfer among the participating atoms .
when sublimation of atoms occurs , point vacancies are created on the surface .
quantitative information of the amount and type of atoms sublimated , as well as point vacancies created on the surface at different annealing temperatures thus can be obtained this way . despite the simplicity of the idea , to the best of our knowledge
we have not found any literature that uses md to simulate such process .
experimentally , if we heat a zno wurtzite surface to an elevated temperature and investigate the resultant surface using photoluminescence ( pl ) measurement , the spectrum should reflect , at least qualitatively , the amount of point vacancies created . based on general considerations , we expect that increase of annealing temperature will create more point vacancies . in this paper , the predictions of md simulation using reaxff are compared against the pl data measured experimentally by our group . in section [ methods ]
we describe the md procedures used to perform the simulations . in section [ experiment ]
we describe the experimental procedure to perform pl measurement on the zno samples . in section [ rnd ] ,
the experimental data and simulation results are discussed , compared and interpreted .
we conclude in section [ conclusion ] .
crystalline zno can exist in various polymorphs which stability is dependant upon the pressure and temperature @xcite .
these polymorphs include wurtzite , zinc blende and rocksalt .
zno in wurtzite structure is the most stable form at room pressure .
the parameters of crystal structure of the wurtzite zno unit cell were shown in table [ znopara ] .
a slab comprising 15 @xmath6 15 @xmath6 3 unit cells was constructed .
the slab was sandwiched between two vacuum layers of thickness 100 @xmath7 , while the thickness of the slab itself was 15 .
the surface area of the slab in the supercell was 1783 @xmath8 .
the thickness of vacuum was chosen such that the interaction between adjacent surfaces of two neighbouring supercells were negligible to not give rise to any significant effect on the simulation outcome .
periodic boundary condition was imposed in all @xmath9- , @xmath10- and @xmath11-directions .
[ znoorig ] shows the supercell of the zno slab . in all simulations
the total number of atoms in the simulation box were maintained at 2700 .
the md simulation throughout this work was carried out using the md code lammps @xcite .
the stepsize throughout the simulation was @xmath12 = 0.5 fs .
the structure was first optimized at 0.1 k using the built - in conjugate gradient minimizer available in the lammps package .
we set the convergence criterion to be 10000 steps ( generally , convergence is achieved in less than 100 steps ) .
the temperature of the system was then heated up to @xmath13 k in 5000 steps .
the system was further equilibrated for 20000 steps at 300 k. at the end of the equilibration at @xmath14 , the temperature was raised to a chosen target temperature , @xmath2 , at a rate of 5@xmath15 k / s .
the evolution of the system at annealing temperature was then followed at constant @xmath2 for a total of @xmath16 = 250000 steps .
the temperature of the system was then quenched from @xmath2 to 0.1 k at a rate of @xmath17 k / s .
a typical temperature vs. step profile is shown in fig .
[ temperature_vs_i ] for annealing temperature @xmath2 = 1000 k. nose - hoover thermostat ( nvt ) was used throughout the simulation to control the temperature .
the damping constant for the thermostat was set to 5 fs .
the total steps in each simulation are dependent upon the target temperature @xmath2 .
the total duration for the simulations range from 315 000 steps ( 0.16 ns ) to 387 000 steps ( 0.20 ns ) .
independent simulation was carried out for each target temperature @xmath2 , ranging from 300 k to 1300 k at an interval of 100 k. atoms from the polar surfaces will leave the surface if the target temperature is sufficiently high . in this case
we have to assure that @xmath16 that defines the length of the temperature plateau is sufficiently long such that atom will stop leaving the surface even if we continue to equilibrate at that annealing temperature .
this precaution is necessary so that the conclusion of the md outcome remains the same should a larger @xmath16 is used instead . after some trial - and - error efforts
, it was found that a value of @xmath18 steps was optimal for the range of temperatures investigated in our work . at the end of the simulation ,
i.e. when the slab has been quenched to 0.1 k from a given target temperature @xmath2 , the type and total number of atoms leaving the surface were quantified . after running a series of simulation for various @xmath2 ,
we would have collected sufficient data to plot the ratio of atoms leaving the surface as a function of annealing temperature .
the reliability of the md results depends crucially on the force field used .
force field based on the reaxff model , first proposed in @xcite , allows bond formation and charge transfer to occur among the interacting atoms in the simulation box . in this work ,
the reaxff for zno is adopted .
the zno reactive force field has been applied to the calculation of atomic vibrational mean square amplitudes for bulk wurtzite - zno for a temperature up to 600 k and found good agreement with experiments @xcite .
it has also been applied to study surface growth mechanism for the wurtzite ( 0 0 0 1 ) surface and water molecule adsorption on stepped zno surfaces @xcite .
the parameters of the reaxff used in this work was the same as that used in @xcite . through reaxff the process of equilibration of partial charge distribution among the atoms
can be revealed in the md simulation . in our simulations , all the atoms in the simulation box were assigned a partial charge of zero initially . while total charge of the system remains zero , partial charges among the atoms will redistribute itself as thermal equilibration process progresses .
we followed the time evolution of partial charge distribution in the slab when the zno is going through the annealing process .
the distributions of the partial charge density in the slab at the end of a md run were obtained by post - processing the md data .
photoluminescence ( pl ) measurement was carried out to justify the defect caused by the annealing procedure .
two set of zno powders were used as starting material in this study .
they were synthesized using french process @xcite , having different major morphological particles : rod ( zno-1 ) and plate structures ( zno-2 ) .
the zno powders were converted into pellets form using the zno agglomeration method , which was introduced in the previous work @xcite .
the zno pellets formed using this technique retained the morphological structure of zno .
x - ray diffraction ( xrd ) analysis had been carried out on the zno pellets and revealed the wurtzite polycrystalline structure of the zno samples ( not shown here ) .
the zno pellets were subsequently annealed in nitrogen rich ambient for an hour with the gas flow regulated at 2.4 l / min .
the annealing process was done in temperature manipulated at 573 k
1073 k. the experimental work was conducted using photoluminescence spectroscopy ( jobin yvon hr 800uv ) with hecd laser 325 nm as excitation source was used to study the luminescence properties of the sample and to indicate the types of recombination center for the semiconducting samples .
the level of defect formation was also determined from the luminescence intensity .
the experimental results are to be discussed in subsection [ comparison ] .
we simulated the annealing of the zno slab for @xmath20 k to @xmath21 k at a 100 k interval . at low temperatures ,
the atoms are observed to execute relatively mild thermal vibration about their equilibrated positions . at higher temperature ,
the vibration become more violent and the configuration of the atoms on both surfaces became more distorted as compared to the low temperature cases .
however , the layered structure formed by the alternating layers of oxygen and zn atoms remains visibly recognizable .
the video of the simulations were visually monitored for any atoms leaving the surfaces of the zno slab .
we observed that o atoms began to leave the ( 0 0 0 1 ) surface , mostly in pair , for simulation running with @xmath2 = 800 k or above .
sublimation of the o atoms occurs only during the temperature plateau phase .
no atom other than the o atoms on the ( 0 0 0 1 ) surface sublimates from the slab .
in particular , no zn atom sublimates from either surfaces of zno . such observation tallies with the experimental measurement that only oxygen vacancies were created but not zn upon heating a zno sample ( see the related discussion in subsection [ comparison ] ) .
[ a4vovera4vs_vs_temp ] shows the ratio of o atoms sublimated to the total o atoms originally lying on the ( 0 0 0 1 ) surface for each target temperature .
no sublimation occurs for @xmath22 k. o atoms sublimate from its polar surface only happened for @xmath23 k or above , with more atoms leaving the ( 0 0 0 1 ) surface at larger @xmath2 .
[ a4vovera4vs_vs_temp ] implies the existence of a threshold temperature , 700 k @xmath24 k , at which sublimation of o atoms from the zno surface begins to occur .
we shall qualitatively compare this prediction against experimentally measured data in subsection [ comparison ] .
it was also observed that most of the sublimated o atoms will leave the ( 0 0 0 1 ) surface in pairs
. a closer investigation of the simulated snapshots revealed that the two oxygen atoms forming a pair during sublimation were neighbors to each other before both left the surface .
however we also observed a tiny portion of sublimated o atoms leaving the surface alone and did not form pairs .
the formation of oxygen atom pairs during sublimation was made possible in the md simulation due to the novel capability of the reactive force field to allow bond formation between neighboring o atoms . from theoretical point of view , surface atoms are under very different conditions than that for the bulk atoms , give rise to the presence of dangling bonds in the former . as a result , the polar surface which consists of o atoms have higher tendency to form bond with a neighboring o atoms from the same surface layer . when sublimated , these dimers form oxygen pairs as observed in the md simulation .
based on our md results , we conclude the following microscopic picture : dimers of o atoms will be formed on the wurtzite ( 0 0 0 1 ) zno surface when it is heated up .
these dimers shall leave the surface in pairs if subsequently sublimated .
the experimental data obtained from pl measurement as described in section [ experiment ] is shown in fig .
[ plspectrum ] .
the origin of the primary peak ( at wavelength 382 nm ) in the pl data is due to the bandgap edge as well as excitonic recombination .
the origin of the broad pl secondary peak , i.e. green emission centered at 536 nm , remains a debatable issue .
it could be interpreted as a reliable indicator of the abundance of point defect present on the surface or otherwise . in either case
, it is reasonable to reckon that the intensity of the green emission is related to the presence of point defects .
the only issue is the quantitative extent of the contribution of the point defect apart from other possible causes . in this paper
we adopt the viewpoint that the intensity of the secondary pl peak is taken as a qualitative indicator of the abundance of point defect present on the surface . such a viewpoint is consistent with the other reports @xcite .
ahn et al . had explained that the broad green peak centered at 536 nm is due to band transition from zinc interstitial ( zni ) to oxygen vacancy ( vo ) defect levels in zno @xcite .
al . reported the band transition from zni to vo level is approximately 2.31 ev ( green emission ) based on the full potential linear muffin - tin orbital method @xcite .
we shall interpret the pl data by making the following assumptions : 1 .
the larger the abundance of oxygen point defects on the surface the higher the intensity of the secondary peak ; 2 .
the primary peak contribution to the oxygen vacancies on the surface is not to be considered in present analysis ( which is still qualitative ) fig .
[ pl_vs_temperature ] shows the intensities of green luminescence ( gl ) at wavelength 512 nm of the zno-1 and zno-2 samples as a function of temperature .
the zno samples at @xmath25 k , @xmath26 573 k , 673 k , 773 k and 873 k display a low intensity .
there is a pronounced increase of gl luminescence at @xmath2 = 973 k and @xmath2 = 1073 k , with the intensity at the latter temperature higher than the former .
when @xmath2 crosses 873 k to 973 k , both zno-1 and zno-2 samples ( with different morphologies ) display the same threshold effect of abrupt increase in the gl intensity .
a temperature threshold effect is clearly being triggered between @xmath2 = 873 k and @xmath2 = 973 k , which is to be compared with @xmath27 800 k predicted by the md simulation .
the predicted threshold temperature is based on idealized model in a md simulation that can not exactly describe what actually happened in a real experiment .
since the parametrisation of the zno reaxff does not include sublimation of o atoms from zno surfaces , the value of @xmath19 from md falls so closely to the experimentally measured range of @xmath28 973 k is a very suggestive finding .
if we accept an uncertainty of @xmath29 k in the threshold temperature , which is reasonably acceptable for a typical md calculation , then the following picture seems appropriate ( at least qualitatively ) : ( _ i _ ) oxygen point defects on the wurtzite zno ( 0 0 0 1 ) surface are created only at or above the threshold annealing temperature of @xmath19 , and ( _ ii _ ) the abundance of oxygen point defects on the surface increases as the annealing temperature @xmath2 increases ( where @xmath3 ) . apart from the pl data as reported above , the md simulation results are also consistent with a previous experiment @xcite which was conducted by two present authors ( c.a.l and s.m ) . in @xcite , two different zno nanostructures ( in wurtzite phase )
were investigated using electron spectroscopy imaging ( esi ) .
one sample was not annealed while the other was annealed to @xmath30c (= 973 k ) under @xmath31-rich condition .
the surfaces of the samples were scanned for zn and o atoms .
the abundance of these atoms on the surfaces could be visually inspected and were characterized by two distinct colors .
the images of the esi measurement on the white zno samples were reproduced in fig .
[ fig : ling ] .
it was found that the density of zn atom count is similar for both annealed and unannealed samples . however , the density of o atom count for the annealed sample has clearly decreased compared to the unannealed sample .
the esi images are consistent with the notion that o atoms will be sublimated at a temperature @xmath32 , while zn atoms will not .
although the zno samples were measured for only two temperatures ( one at ambient and the other at 973 k ) in @xcite , the data is ubiquitously consistent with the present md results .
the qualitative agreement between the md results and the two experimental measurements discussed above is encouraging .
it provides a motivation to capitalize the md simulation as a reliable means to derive useful information of the detailed mechanism of the zno surface undergoing heating at the atomistic level , which is otherwise difficult to obtain via experimental approach alone .
the time evolution of the partial charge distribution among the atoms in the zno sample can be followed throughout a md course ( during which the slab is undergoing a temperature history as in fig .
[ temperature_vs_i ] ) . as an illustration , figs .
[ partialcharge_vs_z300 ] , [ partialcharge_vs_z1300 ] displays the snapshots of the partial charge distribution in the slab at various steps during the md simulation at target temperatures @xmath33 300 k and @xmath33 1300 k respectively . in these figures ,
the blobs centered around a given @xmath11-position ( representing the vertical distance from the ( 0 0 0 1 ) surface ) are made up of discrete dots , each represents the partial charges of an atom at the depth @xmath11 .
partial charge with a positive ( negative ) value is associated with zn ( oxygen ) atom . the partial charge distribution for the unannealed slab ( i.e. ( a ) in fig . [ partialcharge_vs_z300 ] and fig .
[ partialcharge_vs_z1300 ] ) displays an oscillation pattern .
the oscillation is due to the fact that the sign of partial charges from the same atom type at a common layer is opposite to that of adjacent layers occupied by the other atom type . in our md simulations
using reaxff model , the partial charge of o atom is negative , while it is positive for zn atom .
the signs are in accordance with what we expected from both atom types based on their known electronegativity .
when probing into the depth of the slab along the @xmath11-direction beginning from the ( 0 0 0 1 ) surface termination , atomic layers with alternating partial charges will be passed through in succession , with the first encountered layer being the negatively charged o layer .
oscillation of the partial charge density in the slab observed here has a strong resemblance to what is known as friedel oscillations which lead to surface charge polarization , a common feature of all surfaces . by following the snapshots
sequentially in time , we can quantitative visualize the information of how the positions and partial charges of the atoms redistribute themselves throughout a md course .
for example , at the beginning , the blobs of points in ( a ) in figs
. [ partialcharge_vs_z300 ] , [ partialcharge_vs_z1300 ] are very concentrated . from this
, it could be inferred that ( _ i _ )
all atoms were arranged in layers according to their respective atom type , and ( _ ii _ ) atoms in the same layer share an approximately same value of partial charges .
atoms in the slab annealed to 300
k vibrate only mildly , as can be inferred from the smaller size of the blobs in ( b ) and ( c ) in fig .
[ partialcharge_vs_z300 ] . at the end of the annealing process at 300 k , the atomic configuration of the slab , as well as the partial charges of the atoms ,
basically did not alter much .
whereas , from fig . [ partialcharge_vs_z1300 ] for slab annealed at 1300 k , atoms vibrate violently about their atomic positions at the temperature plateau ( blobs in ( b ) and ( c ) are more dispersed than in the 300 k case ) . at the end of the annealing process , the atomic positions became much distorted .
the partial charges of the same atom type in a given layer spread to a broader value ( as can be inferred from a large vertical spread in the blobs ) . comparing graphs ( a ) and ( d ) in fig . [ partialcharge_vs_z1300 ]
, the distortion resulted from the annealing process at @xmath34 k is particularly prominent near the ( 0 0 0 @xmath0 ) side of the surface , where a large spread in partial charges and locations of the zn atoms are observed .
in general we found that the ( 0 0 0 @xmath0 ) end of the slab crumpled at the end of the annealing process for @xmath2 equal or larger than 300 k. fig .
[ partialcharge_vs_zlast ] displays the partial charge distributions at the end of an md run for eight annealing temperatures ranging from @xmath35 k to 1100 k. these figures are basically the compilation of stage ( d ) in figs . [ partialcharge_vs_z300 ] , [ partialcharge_vs_z1300 ] for eight different @xmath2 ranging from @xmath35 k to 1100 k. a qualitative change in the partial charge distribution pattern between @xmath33 700 k and @xmath33 800 k was observed . for @xmath36 700 k ,
the partial charge point were grouped in relatively concentrated blobs , which were slightly flattened in the horizontal direction . upon crossing the temperature at 800
k , the points of partial charges in these blobs disperse to a much larger extent along the @xmath11-axis .
to illustrate the temperature - induced abrupt change of partial charge distribution from a slightly different perspective , the density of net partial charges at a depth of @xmath11 from the ( 0 0 0 1 ) surface @xmath37 , which was derived from the partial charge distribution vs. @xmath11 data , is shown in fig .
[ qdensityprofile ] for annealing temperatures @xmath38 k. the density of net partial charges at a depth @xmath11 , @xmath37 , is defined as the sum of all partial charges in a volume element of @xmath39 , where @xmath40 the surface area of the slab .
the variation of @xmath37 at different depth for a few @xmath2 shown in fig .
[ qdensityprofile ] displays a qualitative change in the density of partial charge profile when @xmath2 crosses 700 k to 800 k. the qualitative change of the partial charge distribution coincides with the onset of o atoms sublimation in this temperature range described in subsection [ sublimation ] . in passing
, we also enumerated the average partial charge of the sublimated o atoms at each temperature , see fig .
[ qv4peratom_vs_t ] .
when the sublimated atoms first appeared at @xmath41 , the average net charge per atom was @xmath42 , and this value increased to @xmath43 at @xmath34 k. the temperature dependence of the average partial charges of the sublimated o atoms appeared to flatten out for @xmath2 beyond 1100 k. when the o atoms were sublimated from the surface , they carry away with them a net negative charge , leaving behind a substrate that has an overall net positive charge on the surface .
we have also repeated the md annealing procedure with different surfaces , i.e. , ( 1 0 0 0 ) and ( @xmath0 0 0 0 ) .
these surfaces were obtained by re - orientating the original unit cell .
the two surfaces in this construction were terminated by oxygen - zinc hybrid atoms .
in fact , most of the low dimensional surfaces of zno crystal phases were terminated by zn or oxygen - zinc hybrid atoms , except the ( 0 0 0 1 ) surface in wurtzite phase ( which forms the oxygen atom terminated polar surface ) .
the ( 1 0 0 0 ) and ( @xmath0 0 0 0 ) are non - polar surfaces . in our effort to anneal the ( 1 0 0 0 ) and ( @xmath0 0 0 0 ) surfaces of the zno slab , we have found no robust sign of sublimation of any atom from the surfaces .
the near absence of sublimation suggests that o atoms only sublimate from the ( 0 0 0 1 ) surface in wurtzite zno upon annealing , while not so from any other zno crystal surfaces terminated by zn or oxygen - zinc hybrid atoms .
our md simulation results suggest the possibility that the zn atom on the surface termination presents a strong attractive force to any o atom attempting to escape the surface via thermal excitation .
however such hypothesis needs to be tested more rigorously by investigating the detailed behavior of the force of the zn atom on any o atom attempting to leave the surface .
we have performed md annealing simulations of a wurtzite zno slab with a thickness of 15 @xmath45 and a surface area of 1783 @xmath8 .
the simulations are performed using zno reaxff , as published by @xcite .
there are two polar surfaces , namely ( 0 0 0 1 ) and ( 0 0 0 @xmath0 ) . the former is o terminated while the latter is zn terminated .
the slab is thermally annealed from 0.1 k in stages up to a target temperature @xmath2 , and then further equilibrated at the constant @xmath2 plateau for sufficiently large number of steps before the temperature is gradually quenched to 0.1 k. the common temperature history experienced by the polar surfaces are shown in fig .
[ temperature_vs_i ] .
we have run md simulations for a range of target temperature from @xmath25 k till 1300 k at a 100 k interval .
our md results show that o atoms on the ( 0 0 0 1 ) surface is sublimated whenever @xmath3 , where 700 k @xmath1 k. otherwise no atom sublimates from either surface . the ratio of o atoms sublimated to the total number of o atoms on the surface increases as @xmath2 ( @xmath3 ) increases .
the existence of a threshold temperature @xmath19 at which o atoms from the ( 0 0 0 1 ) surface begins to sublimate qualitatively agrees with our pl data ( obtained experimentally ) and a previous qualitative observation @xcite based on esi imaging .
the md simulation results also show that the sublimated o atoms do so in pairs . due to the geometrically asymmetric condition
, atoms at the surface are expected to form dimers .
the o pairs seen leaving the surface in the md sublimation originate from these dimers .
we have also investigated the partial charge density of the slab as a function of depth from the surface , @xmath11 , after each slab has gone through an annealing history as depicted in fig .
[ temperature_vs_i ] .
the oscillatory form of the charge density of the slab resembles friedel oscillations which are expected for the charge density near a surface .
a qualitative change in the partial charge distribution pattern was observed while the annealing temperature crosses @xmath46 k to @xmath47 k. md annealing performed on the ( 1 0 0 0 ) and ( @xmath0 0 0 0 ) surfaces , which are terminated by oxygen - zinc hybrid atoms , did not produce any sublimation of either o or zn atoms .
this suggests that thermally - driven sublimation of atoms only occurs in the o - terminated ( 0 0 0 1 ) surface , whereas no sublimation of either atom type would occur in the other surfaces ( which are mostly oxygen - zinc hybrid or zn - terminated ) presumably due to a strong attractive force exerted by the zn atoms at these surfaces .
t. l. yoon wishes to acknowledge the support of ( 1 ) frgs grant fasa 2/2013 by the ministry of higher education of malaysia ( 203/ pfizik/6711348 ) , ( 2 ) usm ru grant ( 1001/psosial/816210 ) .
we also acknowledge the financial support from an ergs grant ( 203/pfizik/6730100 ) from the malaysian government . | the effect of thermally annealing a slab of wurtzite zno , terminated by two surfaces , ( 0 0 0 1 ) ( which is oxygen - terminated ) and ( 0 0 0 @xmath0 ) ( which is zn - terminated ) , is investigated via molecular dynamics simulation using reactive force field ( reaxff ) .
we found that upon heating at or beyond a threshold temperature 700 k @xmath1 k , surface oxygen atoms begin to sublimate from the ( 0 0 0 1 ) surface , meanwhile no atom leaves the ( 0 0 0 @xmath0 ) surface . the ratio of oxygen leaving the surface at a given temperature @xmath2 increases as @xmath2 increases ( for @xmath3 ) .
a photoluminescence ( pl ) measurement has been carried out to assess the effect of annealing wurtzite zno surfaces .
the relative luminescence intensity of the secondary peak in the pl data , interpreted as a measurement of amount of vacancies on the sample surfaces , qualitatively agrees with the threshold behavior as found in the md simulations .
our simulations have also revealed the formation of oxygen dimers on the surface and evolution of partial charge distribution in the slab during the annealing process .
we have also simulated the annealing on the ( 1 0 0 0 ) and ( @xmath0 0 0 0 ) surfaces but found no robust sign of sublimation .
the thermal behavior of a zno slab revealed by our md simulation based on the reaxff is consistent with experimental observations .
molecular dynamics , reactive force field , annealing , zno surface |
antiferromagnets ( afms ) show great potential to replace ferromagnets ( fms ) in spintronic applications @xcite . compared with fms , afms
have the advantages of much faster spin dynamics @xcite , more stability against charge and external field perturbations @xcite , and no stray field @xcite .
however , since the afm spins align in alternating directions of magnetic moments on individual atoms , the resulting zero net magnetization makes hard to control afm magnetic moments .
recently , there have been several reports regarding the control of afm moments by applying an electronic current in afm films @xcite and fm / afm bilayers @xcite , by field cooling ( fc ) @xcite and by applying an external field via the exchange - spring effect @xcite .
these studies demonstrated that the afm moments can be controlled and detected using electronic transport measurements without the need for large - scale facilities such as synchrotron and neutron facilities @xcite .
afm rotation is of interest because a more than 100% spin - valve - like signal has been achieved in tunneling anisotropic magnetoresistance ( tamr ) stacks by controlling the afm spin configuration via the exchange - spring effect of fm on afm @xcite .
the exchange coupling has been widely used in spintronic devices such as spin - valve - type magnetic memory devices to pin the fm magnetization @xcite .
in contrast , tamr utilizes the rotating afm exchange - coupled to fm @xcite .
the rotating afm can be linked to the shift of hysteresis loops ( exchange bias ) and broadening of the coercivity in magnetization measurements @xcite .
although several studies have been reported regarding the rotating afm @xcite , almost all of the studies have been performed on polycrystalline stacks using afm for irmn .
since the afm moments rotate with exchange - coupled fm , the afm rotation behavior is expected to be affected by the fm magnetization switching process .
thus , the effect of fm magnetocrystalline anisotropy resulting from the full epitaxial growth is more interesting .
in addition , all the studies have been performed using typical @xmath1 metal fms such as nife and co. similar to successful studies on giant magnetoresistance ( gmr ) and tunneling magnetoresistance ( tmr ) @xcite , there is a clear need to study high - quality heterostructures with more advanced compounds such as those with high - spin polarization needed for spintronic devices @xcite . for the advanced materials , we focused on the heusler compound fe@xmath0crsi ( fcs ) , which is theoretically expected to be half - metallic fm @xcite .
the four - fold magnetocrystalline anisotropy constant of fcs is known to be 266 j / m@xmath2 @xcite .
in addition , since only fully epitaxial stacks effectively provide the advanced properties , the heusler compound ru@xmath0mnge ( rmg ) was selected for afm ; we succeeded in growing fully epitaxial fcs / rmg bilayers @xcite .
rmg exhibits the highest n@xmath3el temperature , @xmath4 k , among heusler compounds @xcite .
in addition , rmg has a nearly half - metallic electronic structure @xcite .
since the tamr depends on spin configuration of afm , such an afm electronic structure is interesting . in this letter
, we systematically studied the magnetic and magnetotransport properties of fcs / rmg bilayers using current - in - plane ( cip ) configurations . the angular - dependent resistance change ( @xmath5 ) ratio and exchange bias ( @xmath6 )
exhibit a similar rmg thickness ( @xmath7 ) dependence , indicating that rmg spin reorientation via fcs rotation is dominant in @xmath5 above a critical thickness ( @xmath8 ) to induce @xmath6 even in cip configurations . above @xmath8 , the magnetoresistance ( mr )
curves along the hard axes of fcs magnetocrystalline anisotropy exhibit a full rotation of afm moments . however , a partial rotation of afm moments is observed along the easy axes , demonstrating the effect of the fm magnetization reversal process on afm rotation .
although previous studies have used current - perpendicular - to - plane configurations , these results indicate that cip magnetotransport measurements in fm / afm bilayer provide a facile approach to detect afm moments and promote their application in afm spintronics .
/@xmath9 scans ( a ) and in - plane @xmath10-scans of fcs / rmg bilayers .
, width=264 ]
rmg / fcs bilayers were deposited on mgo ( 001 ) substrates by dc magnetron sputtering at a base pressure of approximately @xmath11 torr .
rmg thin films were deposited at a substrate temperature of @xmath12 @xmath13c and then cooled to room temperature .
next , fcs was deposited on rmg at room temperature .
after the fcs was deposited , the fcs / rmg bilayers were annealed at 500 @xmath13c for 30 minutes to achieve @xmath14 ordering of the fcs .
in addition , we also deposited fcs at @xmath12 @xmath13c .
the results were the same in both cases .
the crystal structure was analyzed using both in - plane and out - of - plane x - ray diffraction ( xrd ) measurements with cu @xmath15 radiation .
the magnetic properties were characterized using vibrating sample magnetometry and superconducting quantum interference device ( squid ) magnetometry .
the magnetotransport measurements were performed using the standard dc four - terminal method in the cip configuration . to induce exchange coupling , the bilayers were annealed at 350 k for the squid measurements and at 375 k for the magnetotransport measurements for 30 minutes with applying field of + 10 koe , and then cooled to @xmath16 k with applying field of + 10 koe @xcite .
xrd patterns of the fcs / rmg bilayers are presented in fig .
[ fig : one](a ) and [ fig : one](b ) . as observed in fig .
[ fig : one](a ) , only the @xmath17 fcs and @xmath17 rmg peak series show bragg peaks in the out - of - plane xrd pattern .
epitaxial growth is confirmed by the in - plane @xmath10-scans presented in fig .
[ fig : one](b ) . both rmg ( 111 ) and fcs ( 111 ) peaks are observed with shifts of 45@xmath13 relative to the mgo ( 111 ) peaks .
these results indicate that their epitaxial relationship is fcs(001)[100]//rmg(001)[100]//mgo(001)[110 ] .
in addition , since ( 111 ) reflection peaks of the heusler alloy originate from superlattice reflections in the @xmath14 ordered structure @xcite , these xrd results indicate that high - quality fcs / rmg bilayers were obtained .
, 15 and 20 nm at @xmath16 k after fc , respectively .
( d ) @xmath18 and @xmath6 as a function of @xmath7 at @xmath16 k. , width=321 ] shows the magnetic hysteresis loops of fcs ( 5 nm)/rmg ( @xmath7 nm ) bilayers measured at @xmath16 k after fc .
the measurements were performed along the easy axis of fcs @xmath19 .
the @xmath20 and 15 nm bilayers exhibit narrow hysteresis loops with a coercive field @xmath18 of approximately 100 oe , which is similar to single - layer fcs films .
the @xmath21 nm bilayers exhibited a much wider hysteresis loop with @xmath18 of approximately 860 oe . on the other hand , the @xmath20 nm bilayer exhibited no hysteresis loop shift , whereas the @xmath22 and 20 nm bilayers exhibited @xmath23 and 155 oe , respectively .
the @xmath7-dependent @xmath18 and @xmath6 results are summarized in fig . [ fig : two](d ) at @xmath16 k. @xmath6 is confirmed at above @xmath24 nm , indicating that @xmath8 is between @xmath20 and 10 nm .
a maximum @xmath6 appears at @xmath21 nm ; then , @xmath6 decreases with increasing @xmath7 . the same @xmath7 thickness dependence was observed at @xmath25 k for a wider @xmath7 range @xcite .
@xmath18 shows no significant thickness dependence below @xmath22 nm ; then a jump is observed at @xmath21 nm .
next , we focused on the @xmath7-dependent magnetotransport properties .
figure [ fig : three](a ) plots the @xmath5 ratio as a function of the relative angle @xmath9 between the current and fm magnetization direction of fcs(5 nm)/rmg(@xmath26 ) bilayers at @xmath16 k under an applied field of + 4 koe after fc . at @xmath27
nm , a typical anisotropic magnetoresistance ( amr ) ratio is confirmed with a negative value of approximately -0.04 @xmath28 .
the negative amr sign may originate from the half - metallic electronic structure of fcs , as discussed in recent theoretical and experimental studies @xcite . at @xmath22
nm , the amplitude of @xmath5 increases , and its angular dependence shifts by approximately 45@xmath13 . at @xmath21
nm , the amplitude of @xmath5 is more than an order of magnitude larger than that of single - layer fcs films .
the @xmath5 ratios with respect to @xmath7 at @xmath16 k after both zero - field cooling ( zfc ) and fc are summarized in fig .
[ fig : three](b ) . after fc ,
the @xmath5 ratios are independent of @xmath7 below @xmath8 . above @xmath8
, the @xmath5 ratios increase at 10 and 15 nm . at @xmath21
nm , the @xmath5 ratio drastically increases and then decreases upon further increasing @xmath7 . these @xmath7-dependent @xmath5 ratios are similar to the exchange bias , as observed in fig .
[ fig : two](d ) .
moreover , as observed in fig .
[ fig : three](b ) , the @xmath5 ratios are enlarged by fc , demonstrating the effect of exchange coupling on the @xmath5 ratio .
note that since the exchange coupling might exist even without fc @xcite , the @xmath5 ratio after zfc is larger than that for single - layer fcs films .
in addition , a relationship is observed between the @xmath5 ratios , the shape of the mr curves and exchange bias .
the mr curves measured as a function of the applied magnetic field are presented in fig .
[ fig : three](c ) .
the mr curves of the single - layer fcs films ( @xmath27 nm @xcite ) are symmetric , which originates from the fcs amr . in contrast , above @xmath8 , the curves of the bilayers differ from typical amr curves .
the @xmath24 and 15 nm bilayers exhibit small asymmetric curves , and the asymmetry increases at @xmath21 and 30 nm . according to previous fm / afm studies @xcite , the asymmetric mr curves originate from the partial rotation of the afm moments due to the applied external field via the fm rotation , whereas the symmetric mr curves originate from the full rotation of the afm moments .
finally , we would like to discuss the origin of the anomalous magnetic properties of the bilayers .
figures [ fig : four](a ) and [ fig : four](b ) compare different measurement conditions ; the sensing current @xmath29 was applied in directions parallel to fcs / rmg [ 100 ] and [ 1 - 10 ] , respectively . as observed in fig .
[ fig : four](a ) for @xmath30 $ ] , asymmetric mr curves are obtained along @xmath31 $ ] ( @xmath32 ) and [ 010 ] ( @xmath33 ) . on the other hand ,
symmetric mr curves are obtained along @xmath34 $ ] ( @xmath35 ) and [ -110 ] ( @xmath36 ) . for @xmath37 $ ] ,
the symmetric and asymmetric relations with respect to crystalline direction do not change ; symmetric mr curves are obtained along @xmath38 $ ] ( @xmath32 ) and [ 110 ] ( @xmath33 ) , and asymmetric mr curves are obtained along @xmath31 $ ] ( @xmath35 ) and [ 0 - 10 ] ( @xmath39 ) .
although fc - direction dependence was performed , no change was observed .
these results indicate that the symmetric and asymmetric curves are not determined by the relative angle @xmath9 between @xmath40 and @xmath29 , indicating that the obtained angular dependences are not typical amr of fm .
then , because the symmetric and asymmetric relations are not changed by the fc directions , the factor governing the angular dependence of the bilayers is not the sensing current or fc directions but the fcs / rmg crystalline direction .
$ ] and @xmath41 $ ] , respectively .
( c , d ) fcs magnetization switching process where the field sweeps along the easy and hard axes , respectively . , width=321 ] one possible cause of the crystalline - direction - dependent afm rotation is the fm magnetization switching process .
fcs has four - fold magnetocrystalline anisotropy , where the easy axes are oriented along @xmath19 and @xmath42 , and the hard axes are oriented along @xmath43 and @xmath44 @xcite .
this property indicates that symmetric mr curves appear along the hard axes of fcs , and asymmetric mr curves appear along the easy axes of fcs .
it is well known that the fm magnetization switching processes with four - fold magnetocrystalline anisotropy differ along the easy and hard axes . as presented in fig .
[ fig : four](c ) , the magnetization rotates by 180@xmath13 along the easy axes . on the other hand , as presented in fig .
[ fig : four](d ) , the magnetization rotates in 3 steps along the hard axes ; ( i ) the magnetization rotates toward the nearest easy axis , ( ii ) the magnetization jumps in a direction close to the other easy axis , and ( iii ) the magnetization finally rotates toward the applied field direction @xcite .
therefore , the magnetization rotates by up to 90@xmath13 along the hard axis .
a possible explanation for the full and partial afm rotations along fcs / rmg @xmath19 and @xmath43 are that the afm can fully follow the fm magnetization switching via the fm rotation when the field sweep is along the hard axis of fm because the magnetization rotates slightly ( up to 90@xmath13 ) . on the other hand ,
the afm can not fully follow when the field sweep is along the easy axis of fm because the magnetization rotates a lot ( 180@xmath13 ) .
these results could provide a route to understanding the afm rotation behavior .
since the mr curves suggest the importance of afm spin configuration , as discussed above , the larger @xmath5 ratio compared with that of only single - layer fcs films can be considered to be due to afm moments . to date , there have been several studies of afm amr due to spin flop @xcite , crystalline amr originating from large anisotropies in the relativistic electronic structure @xcite and afm spin configurations with respect to current direction @xcite . in the crystalline amr study ,
the afm spins were reoriented by applying magnetic fields via the exchange spring effect @xcite , which is the same condition as that in our study . in addition,the @xmath5 ratio
was obtained under @xmath45 koe with small angle steps ( @xmath4615@xmath13 ) , indicating that the afm moments can fully follow the fm magnetization rotation as mentioned above .
the afm spin reorientation is reinforced by the delay of angular dependence due to the exchange - spring effect @xcite , as shown in fig .
[ fig : four](b ) . then , the resistance is higher for the configuration of afm moments aligned along [ 110 ] and [ -1 - 10 ] than for the configuration of afm moments aligned along [ 1 - 10 ] and [ -110 ] .
these resistance changes due to the afm spin direction have been reported in fm / afm bilayers @xcite . moreover ,
the asymmetric mr curves were transformed into symmetric mr curves with increasing temperature .
therefore , we conclude that the obtained larger @xmath5 ratio compared with that of only single - layer fcs films originates from the reorientation of the afm moments by applying magnetic fields via fm rotation
. the obtained @xmath5 ratio of approximately 5.9 @xmath28 is much larger than other afm amr ratios for sr@xmath0iro@xmath47 of approximately 1 % and mnte of approximately 1.6 % .
this result might be related to either the nearly half - metallic rmg electronic structure or the electronic structures of both rmg and fcs .
we performed a magnetotransport study of fcs / rmg bilayers to clarify the afm rotation behavior .
in addition to the same @xmath7 thickness dependence of the magnitude of the @xmath5 ratio and exchange bias , the mr curves changed from symmetric to asymmetric based on @xmath7 in response to the direction of fcs magnetocrystalline anisotroy .
the maximum @xmath5 ratio of the bilayers was more than an order of magnitude larger than that of single - layer fcs films due to the reorientation of afm moments via the fm rotation .
we also observed that the afm moments could fully rotate when the field sweep was along the hard axes of fcs but could not fully rotate along the easy axes of fcs , demonstrating the impact of fcs magnetocrystalline anisotropy to govern the afm rotation .
these results provide profound insights into the control of the afm moments and promote the application of afm in spintronics .
the authors gretefully acknowledge m. kuwahara and k. saitoh for their invaluable support .
part of this work was supported by the japan society for the promotion of science ( jsps ) program for advancing strategic international networks to accelerate the circulation of talented researchers .
99 h. v. gomonay and v. m. loktev , low temp . phys . *
40 * , 17 ( 2014 ) .
j. zelezn , h. gao , k. vborn , j. zemen , j. masek , a. manchon , j. wunderlich , j. sinova , and t. jungwirth , phys .
rev . lett . * 113 * , 157201 ( 2014 ) .
a. v. kimel , a. kirilyuk , a. tsvetkov , r. v pisarev , and t. rasing , nature * 429 * , 850 ( 2004 ) .
d. kriegner , k. vborn , k. olejnk , h. reichlov , v. novk , x. mart , j. gazquez , v. saidl , p. nmec , v. v. volobuev , g. springholz , v. hol , and t. jungwirth , nat . commun .
* 7 * , 11623 ( 2016 ) .
s. loth , s. baumann , c. p. lutz , d. m. eigler , and a. j. heinrich , science * 335 * , 196 ( 2012 ) .
d. petti , e. albisetti , h. reichlov , j. gazquez , m. varela , m. molina - ruiz , a. f. lopeanda , k. olejnk , v. novk , i. fina , b. dkhil , j. hayakawa , x. mart , j. wunderlich , t. jungwirth , and r. bertacco , appl .
* 102 * , 192404 ( 2013 ) .
p. wadley , b. howells , j. elezn , c. andrews , v. hills , r. p. campion , v. novk , k. olejnik , f. maccherozzi , s. s. dhesi , s. y. martin , t. wagner , j. wunderlich , f. freimuth , y. mokrousov , j. kune , j. s. chauhan , m. j. grzybowski , a. w. rushforth , k. w. edmonds , b. l. gallagher , and t. jungwirth , science * 351 * , 587 ( 2016 ) .
h. sakakibara , h. ando , y. kuroki , s. kawai , k. ueda , and h. asano , j. appl .
phys . * 117 * , 17d725 ( 2015 ) .
tang , h .- w .
zhang , h. su , z .- y .
zhong , and y .- l .
jing , appl .
. lett . * 91 * , 122504 ( 2007 ) .
x. mart , i. fina , c. frontera , j. liu , p. wadley , q. he , r. j. paull , j. d. clarkson , j. kudrnovsk , i. turek , j. kunes , d. yi , j .- h .
chu , c. t. nelson , l. you , e. arenholz , s. salahuddin , j. fontcuberta , t. jungwirth , and r. ramesh , nat .
* 13 * , 367 ( 2014 ) .
a. scholl , m. liberati , e. arenholz , h. ohldag , and j. sthr , phys .
lett . * 92 * , 247201 ( 2004 ) . b. g. park , j. wunderlich , x. mart , v. hol , y. kurosaki , m. yamada , h. yamamoto , a. nishide , j. hayakawa , h. takahashi , a. b. shick , and t. jungwirth , nat . mater . * 10 * , 347 ( 2011 ) . i. fina , x. mart , d. yi , j. liu , j. h. chu , c. rayan - serrao , s. suresha , a. b. shick , j. elezn , t. jungwirth , j. fontcuberta , and r. ramesh , nat . commun . * 5 * , 4671 ( 2014 ) .
x. mart , b. g. park , j. wunderlich , h. reichlov , y. kurosaki , m. yamada , h. yamamoto , a. nishide , j. hayakawa , h. takahashi , and t. jungwirth , phys .
* 108 * , 017201 ( 2012 ) .
s. a. wolf , d. d. awschalom , r. a. buhrman , j. m. daughton , s. von moln@xmath48r , m. l. roukes , a. y. chtchelkanova , and d. m. treger , science * 294 * , 1488 ( 2001 ) .
j. kerman , science * 308 * , 508 ( 2005 ) .
j. geshev , l. g. pereira , and j. e. schmidt , phys .
b * 66 * , 134432 ( 2002 ) .
j. wu , j. s. park , w. kim , e. arenholz , m. liberati , a. scholl , y. z. wu , c. hwang , and z. q. qiu , phys .
* 104 * , 217204 ( 2010 ) .
y. y. wang , c. song , b. cui , g. y. wang , f. zeng , and f. pan , phys .
. lett . * 109 * , 137201 ( 2012 ) .
x. zhou , l. ma , z. shi , w. j. fan , r. f. l. evans , j .-
zheng , r. w. chantrell , s. mangin , h. w. zhang , and s. m. zhou , sci .
rep . * 5 * , 9183 ( 2015 ) .
y. sakuraba , k. izumi , t. iwase , s. bosu , k. saito , k. takanashi , y. miura , k. futatsukawa , k. abe , and m. shirai , phys .
b * 82 * , 094444 ( 2010 ) . h. x. liu , y. honda , t. taira , k. i. matsuda , m. arita , t. uemura , and m. yamamoto , appl . phys . lett . * 101 * , 132418 ( 2012 ) .
m. jourdan , j. min@xmath49r , j. braun , a. kronenberg , s. chadov , b. balke , a. gloskovskii , m. kolbe , h. j. elmers , g. sch@xmath50nhense , h. ebert , c. felser , and m. kl@xmath51ui , nat .
commun . * 5 * , 3974 ( 2014 ) . b. a. hamad , eur . phys . j. b * 80 * , 11 ( 2011 ) .
t. miyawaki , m. foerster , s. finizio , c. a. f. vaz , m .- a .
mawass , k. inagaki , n. fukatani , l. le guyader , f. nolting , k. ueda , h. asano , and m. kl@xmath51ui , j. appl . phys . * 114 * , 073905 ( 2013 )
. n. fukatani , k. inagaki , t. miyawaki , k. ueda , and h. asano , j. appl . phys . * 113 * , 17c103 ( 2013 ) .
n. fukatani , h. fujita , t. miyawaki , k. ueda , and h. asano , ieee trans .
magn . * 48 * , 3211 ( 2012 ) . s. ishida , s. kashiwagi , s. fujii , and s. asano , physica b * 210 * , 140 ( 1995 ) .
y. sakuraba , j. nakata , m. oogane , h. kubota , y. ando , a. sakuma , and t. miyazaki , jpn . j. appl
. phys . * 44 * , 6535 ( 2005 ) .
see supplemental material at . for the exchange bias at @xmath25 k. s. kokado , m. tsunoda , k. harigaya , and a. sakuma , j. phys .
japan * 81 * , 024705 ( 2012 ) .
y. sakuraba , s. kokado , y. hirayama , t. furubayashi , h. sukegawa , s. li , y. k. takahashi , and k. hono , appl .
. lett . * 104 * , 172407 ( 2014 ) .
s. anandakumar , v. s. rani , s. oh , and c. kim , thin solid films * 519 * , 1020 ( 2010 )
. s. br@xmath52ck , j. sort , v. baltz , s. suri@xmath53ach , j. s. mu@xmath53 oz , b. dieny , m. d. bar , and j. nogus , adv . mater . * 17 * , 2978 ( 2005 ) .
since single - layer fcs ( 5 nm ) films did not show clear mr curves , we used the mr curve of fcs ( 200 nm ) . on the other hand , since fcs shows thickness dependent amr ratio , we used amr curves of fcs ( 5 nm ) .
the thickness dependent amr ratio has been reported in single - layer epitaxial fe films ;
m. tondra , d. k. lottis , k. t. riggs , y. chen , e. d. dahlberg , and g. a. prinz , j. appl . phys . *
73 * , 6393 ( 1993 ) .
t. hajiri , t. yoshida , s. jaiswal , m. filianina , b. borie , h. ando , h. asano , h. zabel , and m. kl@xmath54ui , phys .
b * 94 * , 184412 ( 2016 ) . | we report the magnetotransport properties of ferromagnet ( fm)/antiferromagnet ( afm ) fe@xmath0crsi
/ ru@xmath0mnge epitaxial bilayers using current - in - plane configurations . above the critical thickness of the ru@xmath0mnge layer to induce exchange bias , symmetric and asymmetric curves
were observed in response to the direction of fm magnetocrystalline anisotropy . because each magnetoresistance curve showed full and partial afm rotation ,
the magnetoresistance curves imply the impact of the fe@xmath0crsi magnetocrystalline anisotropy to govern the afm rotation . the maximum magnitude of the angular - dependent resistance - change ratio of the bilayers is more than an order of magnitude larger than that of single - layer fe@xmath0crsi films , resulting from the reorientation of afm spins via the fm rotation .
these results highlight the essential role of controlling the afm rotation and reveal a facile approach to detect the afm moment even in current - in - plane configurations in fm / afm bilayers . |
because of the large number of relatively small particles that compose a granular system , granular media tends to act like a collection of microscopic molecules .
however , individual grains in a granular system have a macroscopic mass .
thus , particle collisions and gravity play a much larger role than in molecular systems , making it impossible to fully explain their mechanics in terms of kinetic theory alone . understanding hard sphere ( hs ) systems
is crucial for analyzing important physical phenomena in our world , such as avalanches , earthquakes , and powder production .
newer models that correct for characteristics such as higher densities and dissipation in granular materials have succeeded in explaining simulated and experimental results that were previously left unsolved . in a paper by hong , [ 1 ] , a method for modeling a granular system was developed by observing the effect of lowering the excitation level of a highly excited system of grains .
the excitation of grains is modeled as a temperature reservoir in contact with the system .
when the kinetic temperature , @xmath5 , is high , particles in the system behave like a gas , where the mean free path is much larger than the particle size .
the particles in this state are free to exchange positions with neighboring particles . at the other extreme , @xmath6 ,
the particles remain motionless at the bottom of the container . in this state ,
the particles behave like a solid , remaining locked in a cage , not free to exchange positions with any neighboring particles . as the temperature
is lowered from the gas - like state , we note that a portion of the particles begins to move to the bottom , forming a solid regime , where a layer of particles is locked in a cage .
as the temperature gets closer to @xmath6 , the number of particles condensing to the bottom increases , causing the solid regime , and hence , the number of condensed layers , to grow .
this observation lead to the derivation of a critical temperature , @xmath3 , at which the solid regime begins to form .
hong was able to determine this critical temperature as a function of the control parameters of the system .
he also showed that there is a way to predict the solid regime as a function of @xmath2 once it has dropped below the critical temperature , @xmath3 . in another paper by quinn et al [ 2 ] , the theory proposed by hong was tested and verified using an event driven(ed ) molecular dynamics simulation code .
the results of the ed simulations were compared with the proposed theory .
hong derived results in both two and three dimensions , choosing a particular correlation function for each . in two dimensions ,
hong chose to use the correlation function proposed by ree and hoover [ 3 ] , expressed as follows : @xmath7 with @xmath8 , @xmath9 , and @xmath10 . using this correlation function , hong derived the following expression , @xmath11 with @xmath12 where @xmath13 @xmath14 @xmath15 @xmath16 and @xmath17 is the density at @xmath18 . furthermore , hong was able to determine the following expressions for the critical temperature @xmath3 , @xmath19 and the number of frozen layers @xmath20 , @xmath21 the theory proposed by hong broke the system of hard spheres into two regions , a solid region and a liquid region .
the density profile proposed by hong was shown by quinn to only be valid for the liquid region of the system .
this is because the correlation function proposed by ree and hoover , was not valid for a high density region of hard spheres .
therefore it makes sense that the density function derived from this correlation function , would be inadequate for modeling the high density of the solid region . in a paper by luding et al , [ 4 ] a new correlation function
was proposed for a two dimensional system .
this correlation function is valid for both high and low density regions of a system of hard spheres . in this paper
, we will derive a new density profile using the correlation function proposed by luding , and compare it to the data produced by event driven molecular dynamics simulations .
we will also derive expressions for the critical temperature , @xmath3 , and the number of frozen layers , @xmath20 .
we will also compare these expressions to the simulated data .
in his paper , hong started with the enskog equation for hard spheres [ 5 ] , @xmath22 where the enskog collisional operator , @xmath23 , is given by @xmath24 @xmath25.\ ] ] from this equation , he derived the density profile used to describe the liquid region of the hard sphere system . the same expression can be derived starting with the pressure difference equation , taught in most introductory physics classes . the pressure difference equation that describes the change of pressure _ p _ over height _
z _ , in terms of gravity _ g _ and the density @xmath26 , @xmath27 as a simple example , we can show how this simple equation can be used to derive the maxwell - boltzmann distribution .
starting with the ideal gas law , @xmath28 where _ k _ is boltzmann s constant and _
t _ is the temperature of the system , one can get the following ideal gas pressure in terms of the density , @xmath26 : @xmath29 substituting this expression into eq.(1 ) and solving for @xmath26 , we obtain the following density profile , @xmath30 this is the maxwell - boltzmann distribution , used to model a gaseous phase in statistical mechanics . since the function only applies for the gaseous phase , it does not accurately fit the liquid - solid transition in our system of hard spheres . from the kinetic theory , we can relate pressure and density , where _ m _ equals the mass of a single particle , @xmath31.\ ] ] @xmath32 is a correlation function that takes into account the probability of a collision occurring among particles . in 2001 ,
stephan luding [ 4 ] proposed the following global pressure equation to combine the low - density pressure and high - density expression : @xmath33 \eqno(2)\ ] ] where @xmath34,\ ] ] @xmath35 and @xmath36 -1\ ] ] @xmath37 is a correlation function , similar to the form proposed by ree and hoover .
the first term in eq.(3 ) is a simpler correlation function , introduced by henderson [ 6,7 ] .
the correlation function is determined via a virial expansion around low densities .
the value of @xmath38 , taken at contact , accounts for the excluded volume effect and the increase of the collision rate with density .
@xmath39 is luding s proposed corrected high - density pressure with the numerically fitted constants @xmath40 , @xmath41 , and @xmath42 . included in the global pressure equation , is the following empirical merging function , @xmath43}\ ] ] with the numerical fitting constant @xmath44 .
unfortunately , this function inhibits our ability to analytically solve for the density function @xmath26 . we can obtain an integral expression for the density profile by substituting eq.(2 ) into the pressure difference equation , eq.(1 ) , and solving for the position @xmath45 .
we obtain : @xmath46 @xmath47 @xmath48+z_o \eqno(5)\ ] ] unfortunately , there is no analytical solution to eq.(5 ) , but using numerical methods , one can obtain the solution , and hence , a single function to completely describe the density profile of a two dimensional system of hard spheres under gravity .
one can define the critical temperature as the point at which the density at the bottom layer , @xmath49 , becomes the closed packed density @xmath50 such that @xmath51 for @xmath52 , a portion of the particles near the bottom settle into their minimum energy positions .
the particles in this region are essentially locked in a small cage , where they are free to slightly wiggle , but are never able to exchange positions with other particles in the system .
hence , they form a crystal - like structure . the maximum density @xmath49 depends on the underlying crystalline structure .
the density of square lattice packing is @xmath53 , and the density of a hexagonal lattice is @xmath54 .
we say that particles become locked in position when the density reaches the square - packed density @xmath55 .
however , we will see that the lattice of the solid regime can fluctuate between square and hexagonal packing .
we can use eq.(5 ) and particle conservation to derive an expression for the temperature , @xmath3 at which the solid region begins to form at the bottom .
we begin with the conservation of particles in a column which can be expressed as follows : @xmath56 where @xmath57 is the height of the column of balls at a particular temperature , @xmath2 .
the height @xmath57 can be rewritten as the number of layers simply by factoring out the diameter of a particle , @xmath0 , such that @xmath58 this allows us to rewrite eq.(6 ) as @xmath59 the variable @xmath60 can be defined as @xmath61 we can now define the constant @xmath62 where @xmath3 is the critical temperature and the constant @xmath63 such that @xmath64 now eq.(5 ) can be rewritten as @xmath65 where the function @xmath66 comes from the upper limit of the integrals in eq.(5 ) , and @xmath67 comes from the lower limit of the integrals in eq.(5 )
. then we can rewrite eq.(8 ) as @xmath68 where @xmath69 . solving eq.(9 ) for @xmath45 yields @xmath70 . \eqno(10)\ ] ] we can now substitute eq.(10 ) into eq.(7 ) to obtain @xmath71 @xmath72d\rho\ ] ] @xmath73\ ] ] @xmath74 .
\eqno(11)\ ] ] however , by our previous definitions , we can state that @xmath75 using eq.(12 ) , we obtain an expression for the critical temperature @xmath3 , @xmath76 where @xmath63 is obtained using the following expression : @xmath77d\rho.\ ] ] using the global equation of state and eq.(5 ) , we calculate the value of @xmath63 to be 26.8097 .
thus , @xmath78 we can now derive an expression for the number of frozen layers , using the critical temperature from eq.(14 ) .
the number of frozen layers , @xmath20 will simply be the total number of layers , @xmath79 , less the number of fluidized layers on the top , @xmath80 .
this gives us the following expression for the number of solid layers : @xmath81 finally we obtain the following expression for the number of layers by simplifying eq.(15 ) : @xmath82 the expression we have derived is exactly the same as that derived by hong in [ 1 ] , with a modified value of @xmath3 , due to the global equation of state .
we have used an event driven(ed ) molecular dynamics code to test the condensation of hard spheres due to gravity .
we refer the readers to references [ 8,9,10 ] for details of the algorithm regarding the collisional dynamics .
this particular code takes into account the rotation of the hard spheres as well as a way to handle the inelastic collapse .
the difference between ed and soft - sphere molecular dynamics codes lies in the way time is advanced during the simulations . instead of using finite time increments to advance the system , the ed algorithm finds and advances the system to the next possible event , usually a collision between particles or between a particle and a wall .
this process advances the system in time , but by different time steps for each event that occurs .
the thermal reservoir of our system was modeled using a white noise driving , first introduced by williams and mackintosh [ 11 ] . in this model , a random velocity kick , @xmath83 ,
is added to each particle s velocity every set interval of time , @xmath84 .
these random velocity kicks are not correlated with each other in any way .
the program allows us to set the temperature of the system by setting an input parameter controlling the width of the gaussian from which the random kicks are drawn . in effect
, the temperature parameter controls the temperature of the reservoir and hence , the kinetic temperature of the system .
note that we are _ not _ driving the system with a bottom wall connected to a temperature reservoir , which is used fairly often as a model for the vibrating bed .
fig.(1 ) displays data from a typical simulation of 1000 particles , with initial layers , @xmath85 layers , radii @xmath86 m , and mass @xmath87 kg .
this data is fit with the three density functions derived previously , the mazwell - boltzmann distribtuion , the enskog approximation using @xmath38 , and eq.(5 ) derived from the global equation of state .
it is clear that the maxwell - boltzmann distribution only works for the very low density region , or the gaseous region as expected , and the enskog approximation breaks down once the particles reach their maximum denisty , or solid - like region .
eq.(5 ) fits the data for all densities , making the global equation of state the most accurate for representing our granular system .
we now present typical configurations as a function of temperature using eq.(5 ) . in figs.(2 ) , ( 3 ) and ( 4 ) there are three representative samples at different temperatures for 1000 particles with initial layer numbers of @xmath88 , of mass @xmath87 kg and diameter @xmath89 m. the initial layer thickness is @xmath85 , and the gravitational constant @xmath90 m / s@xmath91 .
seen in fig.(2 ) , at a high temperature ( @xmath92 and @xmath93 ) , all the particles are fluidized and dynamically active .
the density profile is fit using the density derived from the global pressure equation , eq.(5 ) .
since the condensation picture is based on elastic hard spheres , we used both a completely elastic system , where one burst of energy is used to start the system , as well as coefficients of restitution that were close to unity in these simulations .
we note , however , that it was shown in [ 4 ] that the presence of dissipation _ does not _ change the condensation picture .
the critical temperature is just shifted linearly on the order of the coefficient of restitution . in our simulations ,
the coefficient of restitution for all particle collisions in our inelastic systems , whether with the walls or other hard spheres , gives an elasticity of @xmath94 . at a lower temperature ( @xmath95 ) , particles begin to condense and form a solid near the bottom , as shown in fig.(3 ) .
note that the particles first form a loose square - packed lattice , as in fig.(3 ) , and then progressively evolve into a more compact hexagonal lattice structure , as shown in fig.(4 ) .
once the compact hexagonal solid is formed , the density at the bottom remains more or less constant at @xmath96 .
this hexagonal structure is then , for the most part , permanently retained . for fig .
( 3 ) and ( 4 ) , the oscillations in the solid regime are real , but they are simply the finite size effect , i.e , the hexagonal packing in a finite lattice has two more particles in alternative layers
. this oscillation would disappear in the thermodynamic limit .
the critical temperature @xmath3 is determined by the temperature at which a square - packed solid is formed at the bottom layer . beyond this temperature ,
the solid regime steadily approaches the close - packed hexagonal structure and once attained is fixed at @xmath97 .
we point out that between @xmath3 , where there is a square - packed lattice , and lower temperatures , where there is a hexagonally packed lattice , particles squeeze themselves together , expelling holes and progressively forming a compact hexagonal solid . recall that the critical temperature , eq.(14 ) , is a function of @xmath98 and @xmath99 .
eq.(14 ) shows that @xmath99 itself is also a function of @xmath98 .
therefore , as the system changes from square - packed to compact hexagonal packing , so do the values of @xmath98 , @xmath99 , and consequently @xmath3 .
this is because as @xmath98 increases as the solid gets more compact , causing @xmath99 to increase and @xmath3 to decrease .
using an event - driven molecular dynamic simulation written by stephan luding , we obtained data for height versus density for a two - dimensional vibrating bed of hard - discs .
table 1 shows what control parameters we used and how we varied them . for each parameter
set , we ran several different simulations at different temperatures .
table 1 : simulation parameters [ cols="^,^,^,^,^",options="header " , ] using the method of reimann sums in a computer program , we numerically calculated the global equation solution from eq.(5 ) .
then we fit the numerical solution for each density profile to obtain the temperature of each system . from fig.(2 ) , ( 3 ) , and ( 4 ) , it is clear that the numerical solution to eq.(5 ) fits the data extremely well .
we note that there are slight density oscillations in the solid regime for the simulated data due to the presence of the bottom wall .
fig.(2 ) , ( 3 ) and ( 4 ) are all examples of completely elastic systems , with no energy dissipation .
fig.(5 ) and ( 6 ) show comparisons between the elastic and inelastic trials . for the inelastic case , fig .
( 6 ) , it is clear that the global equation still fits the density profile nicely in the liquid - solid regime .
however , the density profile tends to underestimate the height at lower densities .
this is because the particles in the fluid regime are active in the sense that they are free to move around , undergoing collisions , while those in the solid regime are confined in a cage , fluctuating around fixed crystalline positions , and do not undergo collisions .
this effect is discussed further in [ 4 ] , where the global equation was initially developed . to calculate the number of frozen layers for each simulation
, we use eq .
( 16 ) , where @xmath2 is calculated numerically in the simulation and @xmath3 is obtained using eq.(14 ) .
we find , as expected , that the @xmath4 decreases with increasing temperature .
this can be seen in fig.(7 ) , which is data for the case of initial layer number @xmath88 , radii @xmath100 and mass @xmath87 kg .
one can see from the linear fit that the data matches the predictions of eq.(16 ) .
the theoretical slope calculated from eq.(16 ) is @xmath101 while the experimental slope obtained from the simulated data shown in fig.(7 ) is @xmath102 .
this is a percent difference of only @xmath103 .
the other systems produced similar results , all within @xmath104 or less of the predicted values .
( 8) provides a snap - shot , where @xmath105 . in this system
, we would expect to find 30.95 frozen layers .
indeed , we can see from the picture that the first 31 layers appear frozen , whereas particles above this layer can move freely .
these results illustrate validity of the predictions obtained by using the global equation of state .
there are points to consider due to the results presented in this paper .
first , we have demonstrated in this paper , that the global equation of state , first presented by luding et al [ 4 ] , does indeed account for the liquid - solid transition which exists in a system of hard spheres .
the profile derived from the equation of state is useful for determining the kinetic temperature of the system , from the configurational statistics .
the breakdown of particle conservation is no longer an issue when looking at the density profile , because the global equation of state , includes those particles frozen at the bottom .
however , we can still use the concept of a critical temperature to derive the number of frozen layers at the bottem , and compare it to what is observed in our simulated data , due to the fact that the density becomes constant as the solid regime becomes hexoganally - close packed .
the region that is close packed is considered the solid regime and can be used to determine the number of frozen layers in the system and in turn , the thickness of boundary layers . since only a fraction of grains are mobilized under shear [ 12,13 ] , and avalanches and many interesting dynamics occur in these thin boundary layers [ 14,15 ] , such a determination should be of technological importance .
second , the existence of a gravity induced liquid - solid transition of hard spheres must have some interesting consequences to higher order kinetic theory , in particular with regard to the dynamic behaviors .
unlike particles in the liquid regime , those particles in the solid regime are largely confined in cages and fluctuate around fixed positions .
their motions resemble lattice vibrations rather than binary collisions , and it may be a little peculiar , albeit not unphysical , to attempt to describe the lattice vibrations using kinetic theory .
if so , such a description must include much more than binary collisions .
however , the global equation of state gives a complete picture of a system in transition , accounting for both the liquid region with binary collisions as well as the solid region with the lattice regions .
the connection between the two regions , is an empirical one derived by luding et al [ 4 ] .
this gives us some insight into the kintetic details of the transition from liquid to solid regimes , and could be studied further .
as discussed in the beginning and demonstrated in this paper , this gravity induced liquid - solid transition is not a peculiar phenomenon associated with the global equation of state , but rather an intrinsic transition inherent in a system where an excluded volume interaction is dominant .
the formation of a solid at the bottom is the appearance of a massive occupied low energy state due to the pauli exclusion principle .
therefore , the bridge between the liquid and solid regimes , as presented by the global equation of state and the upward spread of the solid regime _ should _ persist because the pauli exclusion principle is in action in real space , even if one may use different approximations [ 16 - 25 ] or may try a different form for the pressure , such as the form suggested by percus - yevick [ 26 ] , and/or in higher order truncation .
it only disappears in the limit when the close volume packing density , @xmath26 becomes one , which is possible only in the case of an ideal appolonian packing [ 27 ] .
one should also further expand the ideas used to derive the global equation of state from two to three dimensions , to see if a similar equation can be derived and tested .
finally , we have shown that the presence of dissipation does not alter the condensation picture at all [ 28,29 ] , _ as long as _ the velocity distribution remains gaussian .
previous experiments [ 30 ] have demonstrated the non - gaussian nature of the velocity distribution , but if the dissipation is small , which is the case for the inelastic simulations carried out in this work , the deviation from gaussian should be small .
we also point out that for hard sphere systems without gravity , there exists no typical energy scale , and thus any transition must be entropy driven , i.e. , there exists no critical temperature , and the phase transition occurs at a critical volume fraction [ 31 ] .
however , for the system considered in this paper , there exists a typical energy associated with the potential energy due to gravity , and thus this transition is not entropy driven , but energy driven , and therefore a critical temperature @xmath3 must exist .
1.0 true cm * acknowledgment * we wish to thank stefan luding for providing us with his ed code and helpful discussions over the course of this work .
[ 1 ] d. c. hong , _ physica a _ * 271 * , 192 ( 1999 ) .
figure 1 : this is a system of 1000 elastic two - dimensional spherical particles with initial layer number @xmath88 , radii @xmath106 m and mass @xmath87 kg .
this displays fits to the data using the maxwell boltzmann distribution , the enskog equation using @xmath37 , and eq.(5 ) , the density profile derived from the global equation of state .
.2 true cm figure 2 : this is a system of 1000 elastic two - dimensional spherical particles with initial layer number @xmath88 , radii @xmath107 m and mass @xmath87 kg .
the temperature of this system is greater than @xmath3 , hence all particles are fluidized .
the red points represents the data points , while the blue is the density profile derived from the global equation of state .
figure 3 : this is a system of 1000 elastic two - dimensional spherical particles with initial layer number @xmath88 , radii @xmath107 m and mass @xmath87 kg .
the temperature of this system is slightly below @xmath3 .
the first few layers at the bottom are beginning to transition to a solid , while the majority of the particles are still fluidized .
the red points represents the data points , while the blue is the density profile derived from the global equation of state .
figure 4 : this is a system of 1000 elastic two - dimensional spherical particles with initial layer number @xmath88 , radii @xmath107 m and mass @xmath87 kg .
the temperature of this system is much less than @xmath3 .
the large number of the bottom layers in the solid - like regime have become hexagonally packed , while a smaller number of particles remain fluidized .
the red points represents the data points , while the blue is the density profile derived from the global equation of state .
figure 5 : this is a system of 1000 elastic two - dimensional spherical particles with initial layer number @xmath88 , radii @xmath107 m and mass @xmath87 kg .
the global fit for the density profile fits the elastic system of particles throughout the entire range of densities .
the red points represents the data points , while the blue is the density profile derived from the global equation of state .
figure 6 : this is a system of 1000 inelastic two - dimensional spherical particles with initial layer number @xmath88 , radii @xmath107 m and mass @xmath87 kg .
the global fit for the inelastic density profile works well for high density but deviates at lower densities .
the red points represents the data points , while the blue is the density profile derived from the global equation of state .
figure 7 : this is a graph of layer number versus temperature for a system of 1000 elastic two - dimensional spherical particles with initial layer number @xmath88 , radii @xmath107 m and mass @xmath108 kg .
as the temperature decreases , more particles become locked in the solid regime .
one can see that the relationship is linear , as is predicted by eq.(16 ) .
figure 8 : this is a snapshot from a system of 1000 elastic two - dimensional spherical particles with initial layer number @xmath88 , radii @xmath107 m and mass @xmath87 kg .
the temperature of this system is much less than @xmath3 .
this simulation corresponds to the simulated data in fig .
( 3 ) . from our derivation @xmath4 for
this system should be 31 layers .
visually , we see that the bottom 31 layers , marked in gray , are indeed in a frozen regime , an area where particles do not switch position with other particles . | using a global equation of state , empirically derived by luding , we accurately model the density profile of a two - dimensional hard sphere system with diameter @xmath0 and mass @xmath1 under gravity with a given temperature @xmath2 [ physica a , 271 , 192 ( 1999 ) ] .
we then compare our solutions to md simulated data . from the density profile
, we can then solve for the critical temperature @xmath3 , which we define as the temperature at which the system begins to condensate .
then , if @xmath2 is below @xmath3 , there is some frozen portion of the system .
we derive a formula for the number of frozen layers @xmath4 , and compare our solution to the number of frozen layers in our simulated data .
pacs number(s ) 64.70.dv , 05.20.dd , 51.10.+y |
the problem is to find a microscopic theory of the second order phase transitions that would allow one to follow a formation of an ordered phase from a disordered phase across an entire critical region continuously .
the microscopic theory should be derived for a given microscopic hamiltonian of a mesoscopic or macroscopic system from the first principles of statistical physics and describe an order parameter , correlations as well as other statistical and thermodynamic quantities . since landau s breakthrough paper of 1937 @xcite
, it is known that the phase transitions , leading to ferromagnetism , bose - einstein condensation ( bec ) , superfluidity , superconductivity , liquid crystal phases , ferroelectricity and so on , originate from restructuring of a many - body system due to spontaneous symmetry breaking .
landau gave a general mean - field theory of spontaneous symmetry breaking .
however , an exact solution for a particular ising model of a magnetic phase transition in a 2-dimensional ( 2d ) lattice of spins , found by onsager @xcite in 1944 and presented in detail by yang @xcite in 1952 , showed that the landau theory was incorrect in the most interesting region surrounding a critical point of a phase transition . according to the ginzburg - levanyuk criterion @xcite , found in 1960 , in the critical region of a system s parameters , the fluctuations of the order parameter are anomalously large , that is they exceed the mean value of the order parameter .
thus , it became clear that critical phenomena are intimately related to the anomalously large fluctuations and could be described only by a much more involved theory .
such a theory should go far beyond the simplified picture of an averaged , mean field .
after the development of powerful quantum - field - theory methods in many - body physics in 1950s and 1960s , that problem was attacked by the best theorists both from the point of view of a general theory and the point of view of various particular phase transitions ( see , for example , @xcite and references therein ) .
they used all the known machinery of the many - body theory , including green s functions , feynman s and matsubara s diagram techniques , wick s theorem , dyson s equation , renormalization group , and so on .
in particular , an important understanding of the universality of phase transitions in the vicinity of the critical point came from the renormalization - group approach @xcite .
however , the latter mostly deals with the thermodynamic - limit quantities and phenomenological scaling and universality , does not fully account for the constraints of a many - body hilbert space ( in particular , due to an employment of the grand - canonical - ensemble or similar approximations ) , and focuses only on a few terms of an intermediate asymptotic expansion at the wings of the critical region , related mainly to the critical exponents .
the renormalization - group approach does not provide a full solution that would be valid both inside the entire critical region and outside it , in the disordered as well as ordered phases .
thus , the current theory of critical phenomena has a well - established microscopic basis and is complete conceptually , but there is the necessity to develop new , regular , and more powerful methods for finding the full solution .
the most striking fact is an absence of a solution to the major problem of how the systems go through a critical point . in general , it is not known for both equilibrium and nonequilibrium phase transitions . here
the principle difficulty lies in an invalidity of the usually used equations for the order parameter , correlation functions and other related quantities in the central part of the critical region . for example , for the bec neither the state of the art equations for the order parameter and the green s functions ( the gross - pitaevskii and beliaev - popov equations ) for the equilibrium case @xcite , nor the quantum kinetic equations for the nonequilibrium case ( see the most advanced analysis in @xcite ) are valid in the critical region . here
we should comment on the terminology , adopted in the literature ( see , e.g. , @xcite ) and employed in the present paper , where the renormalization group approach is named phenomenological to underline its concept of the effective hamiltonians and effective fields .
the latter describe only the long - range correlations in the system at the cost of a rescaling of the short - range fluctuations .
in fact , that rescaling ( i.e. , averaging or course - graining ) procedure constitutes a subtle , often not fully justified and mathematically controlled , approximation in that theory . the exact approach , which takes care of the fields and fluctuations at all scales , is named microscopic .
of course , both approaches constitute the microscopic theory of phase transitions and are based on the same basic principles of the quantum or classical statistical physics of many - body systems , governed by some microscopic interaction hamiltonian .
that terminology is not perfect , but it catches the fact that the renormalization group starts with an approximation of the system and then attempts to find the solution , which could represent only some intermediate asymptotics of the exact full solution . on the contrary ,
the exact approach starts with the rigorous equations , relations and solutions for a mesoscopic , finite system and then analyzes them .
it yields , in principle , more detailed and full information on the critical phenomena and critical functions .
in particular , it goes beyond the thermodynamic - limit critical exponents or intermediate power - law asymptotics of the critical - region wings .
even the exact , nonperturbative renormalization group equations ( for a review , see @xcite ) , including the ones derived by wilson @xcite via an @xmath0-expansion of a space dimensionality @xmath1 for a @xmath2-model , are aimed to approximate the original microscopic hamiltonian by an effective hamiltonian for a large - scale part of a field .
moreover , these exact equations are very difficult to implement due to their complexity . when wilson tried to show `` that the exact renormalization group equations are not hopelessly intractable functional equations '' @xcite , he still substituted them with an approximate recursion formula which is mathematically uncontrollable , but yields the first few orders of the @xmath0-expansion .
so , one must perform approximations and/or truncations , such as in a well - developed method of an effective average action @xcite . in fact , a number of approximations for the renormalization are introduced , including a neglect method , decimation and other real - space renormalization schemes as well as field - theoretical techniques .
however , their rigorous relation to a full exact solution of the mesoscopic problem is still not completely understood .
simultaneously , tremendous effort was spent to study the problem of phase transitions experimentally and numerically in a countless number of various particular many - body systems , including macroscopic and mesoscopic systems , condensed matter , trapped gases , cosmology , relativistic fields , high energy particles , etc .
there is hardly any other problem in physics that attracted such an enormous amount of interest and effort .
suffice it to say that more than thirty nobel prize winners received their nobel awards for the studies related to phase transitions and spontaneous symmetry breaking , including onnes ( 1913 ) , landau ( 1962 ) , feynman ( 1965 ) , onsager ( 1968 ) , n@xmath3el ( 1970 ) , bardeen ( 1972 ) , cooper ( 1972 ) , schrieffer ( 1972 ) , josephson ( 1973 ) , anderson ( 1977 ) , mott ( 1977 ) , prigogine ( 1977 ) , kapitza ( 1978 ) , weinberg ( 1979 ) , glashow ( 1979 ) , salam ( 1979 ) , wilson ( 1982 ) , bednorz ( 1987 ) , m@xmath4ller ( 1987 ) , de gennes ( 1991 ) , lee ( 1996 ) , osheroff ( 1996 ) , richardson ( 1996 ) , cornell ( 2001 ) , ketterle ( 2001 ) , weiman ( 2001 ) , abrikosov ( 2003 ) , ginzburg ( 2003 ) , leggett ( 2003 ) , nambu ( 2008 ) , kobayashi(2008 ) , maskawa ( 2008 ) , englert ( 2013 ) , higgs ( 2013 ) .
hundreds of books and reviews as well as many thousands papers have been devoted to this field of research .
why is the problem of the critical region so important for the theoretical physics ?
an answer follows from a fact that in all other regions of parameters the macroscopic systems show a routine behavior with the normal , gaussian thermodynamic fluctuations , described either by classical or quantum statistical mechanics .
the former was understood in the beginning of the twentieth century due to the works by boltzmann , gibbs et al .
the latter was understood by 1950s after the development of quantum mechanics ( by 1930s ) and quantum field theory ( by 1950s ) .
many - body systems can acquire novel nontrivial properties only in these critical regions , where the microscopic interaction between particles results in a system s self - organization into an ordered phase .
that process restructures the system s configuration and breaks a symmetry , generally , via an internal instability .
thus , a solution to the problem of finding a microscopic theory of phase transitions in the critical region became the major key to progress in our understanding of the macroscopic and mesoscopic systems .
all real systems in the experiments and applications , especially in the modern nanotechnologies , are finite .
their critical regions are also finite and constitute , in fact , the most interesting and important regions of parameters .
resolving the structure of the critical region is equivalent to solving the problem for a mesoscopic ( large , but finite ) system .
the critical region can not be treated as just a discontinuity point , as it was assumed in a standard , bulk - limit analysis of the thermodynamic fluctuations and the proofs of an equivalence of different gibbs ensembles ( see @xcite and references therein ) .
a fact , that the singularities are rounded off in the mesoscopic systems , is well known since the ehrenfest s criticism of the einstein s theory of a phase transition in a gas of bose particles and is relevant to the numerous , especially modern experiments and simulations .
its proper description requires the theoretical , computational tools beyond a thermodynamic - limit analysis .
moreover , the solution of the mesoscopic problem is needed for the computation of the thermodynamic - limit critical functions in a close vicinity of the critical point .
the latter functions describe a continuous transition through the critical point and have a self - similar scaling , different from the power - law scaling at the wings of the critical region .
the rigorous results of the presented in @xcite and in this paper microscopic theory of critical phenomena provide such computational tools and complement the renormalization - group finite - size scaling approach .
moreover , the solution of the problem for the equilibrium systems , which is a subject of the present paper , is crucially important also for the theory of more complicated and diverse nonequilibrium phase transitions .
this theory should describe the formation of an ordered phase in various nonequilibrium processes and is even less studied than the equilibrium one ( see , e.g. , @xcite ) .
unfortunately , despite enormous effort , interest , and importance , the formulated above problem of finding a full microscopic theory of phase transitions has not been solved .
such a theory , that should start from a microscopic hamiltonian and derive the closed universal equations capable to provide the full solution connecting the asymptotics of the ordered and disordered phases across the entire critical region , has not been found so far even for any one of the numerous types of phase transitions .
recently we found a general solution to this problem @xcite . in the present paper
, we formulate that microscopic theory for two cases : bose - einstein condensation in an interacting gas ( sect .
3 ) and ferromagnetism in a lattice of spins , interacting via a heisenberg or ising hamiltonian ( sect . 4 - 6 ) .
we derive the fundamental equations for the order parameter and green s functions , which are valid not only in the fully ordered , low - temperature phase , but also in the entire critical region and in the disordered phase .
these equations allow one to resolve a fine structure of the system s statistics and thermodynamics near a critical point both for the mesoscopic and macroscopic systems .
also , they allow one to find the asymptotics at the wings of the critical region and , in particular , the critical exponents and the critical functions . except for some degenerate cases , like a bec in an ideal gas , a generic process of symmetry breaking is related to an instability in a many - body system that originates from a microscopic interaction between individual particles .
the instability starts from the spontaneous or thermal fluctuations , leads to a growth of the order parameter to a macroscopic level , and ends with a formation of an ordered equilibrium phase with a broken symmetry due to a nonlinear saturation of the instability .
the complex nature of critical phenomena in phase transitions is not reduced even to an interplay of the many - body , instability and nonlinear phenomena .
those processes are subject to a whole set of the simultaneously present and equally important factors , which should be properly taken into account in a correct microscopic theory .
first of all , there are the constraints of a many - body hilbert space
. these fundamental constraints could originate from some local properties of the individual particles or from some nonlocal , global properties of a system s configuration .
( do not mix the term `` constraints of the hilbert space '' with the external parameters or fields in a hamiltonian , like a trap potential @xmath5 or an external magnetic field @xmath6 , which sometimes are also referred to as constraints in the literature @xcite . )
their relation with the integrals of motion , prescribed by a broken symmetry in virtue of a noether s theorem , is also important .
an example of local constraints is given by the occupation numbers @xmath7 of the spin bosons , constituting the spins in the lattice of a ferromagnet .
according to holstein and primakoff @xcite , they are constrained to an interval of values @xmath8 , where @xmath9 is the spin value .
an example of global constraints is a particle - number constraint for a system of @xmath10 particles in a trap . that constraint is responsible , via a related constraint - cutoff mechanism , for the very existence of the bec phase transition and its nonanalytical features @xcite .
the latter implies an insufficiency of a grand - canonical - ensemble approximation , which is incorrect in the critical region @xcite because of averaging over the systems with different numbers of particles , both below and above the critical point . that averaging over the condensed and noncondensed phases at the same time implies a relative error on the order of unity for the critical functions
this is well known , but it is very hard technically to go on with the canonical ensemble .
again , the main problem is a computational , theoretical one .
namely , it is to find a rigorous solution , compatible with the particle - number constraint .
moreover , there is a conceptual problem behind the deviation of the grand - canonical - ensemble approximation from the canonical - ensemble result .
namely , a failure of the grand - canonical - ensemble approximation in the critical region persists even in the thermodynamic limit of a macroscopically large system .
it is demonstrated in fig .
[ fig : c ] for the specific heat of an ideal bose gas in the box and isotropic harmonic traps .
the exact canonical - ensemble and grand - canonical - ensemble specific heats , shown in fig .
[ fig : c ] , are known from an analytical solution for the statistics and thermodynamics of an ideal - gas bec in arbitrary trap .
it was recently found in @xcite and is indistinguishable from the numerically calculated curves in the critical region .
it means that the grand - canonical - ensemble approximation does not allow one to correctly describe a continuous phase transition through the critical point .
that fact is not usually appreciated , probably , in view of an absence of any alternative theoretical tools to compute the result within the canonical ensemble .
the present theory cures that drawback .
note that , even in the thermodynamic limit , the critical region has an order - of - unity width in terms of a self - similar variable @xmath11 .
the latter represents a deviation from the bec critical point in terms of a difference between the number of trapped particles @xmath10 and its critical value @xmath12 , scaled by a dispersion of fluctuations in the number of condensed particles @xmath13 . that variable resolves a fine universal structure of all thermodynamic and statistical critical functions in the critical region .
this structure has a well - defined , smooth limiting form , despite a fact that a size of the critical region in terms of an absolute temperature @xmath14 is shrinking to zero in the thermodynamic limit .
particles , confined in the 3d box with zero dirichlet boundary conditions ( right ) and in the isotropic harmonic 3d trap ( left ) .
the upper plots show the specific heat @xmath15 versus a deviation of temperature from its critical value @xmath16 .
the lower plots show a critical function @xmath17 $ ] that is a deviation of the specific heat from its value at zero chemical potential scaled by a suitable factor @xmath18 , making variation of @xmath19 within a critical region on the order of unity .
the lower plots yield also a universal structure of the specific heat in the thermodynamic limit , since the critical function @xmath20 quickly reaches a limiting form ( i.e. , stops changing ) in that limit due to a use of a self - similar variable @xmath21 .
a thermodynamic limit of the grand - canonical - ensemble ansatz in the usually used approximation of continuous spectrum is shown on the upper plots by the dotted lines and is even more off the exact result.,width=309 ] an absence of an acceptable diagram technique for the constrained many - body systems ( in particular , for the canonical ensemble ) , which has to be used instead of the standard , unconstrained diagram technique based on the grand canonical ensemble , partly explains why the microscopic theory of phase transitions was not found for such a long time . despite many attempts @xcite ,
the required constrained diagram technique has not previously been found .
second , one has to solve the problem for a finite system with a mesoscopic ( i.e. , large , but finite ) number of particles @xmath10 .
it is necessary in order to correctly calculate an anomalously large contribution of the lowest energy levels to the critical fluctuations and to avoid the infrared divergences of the standard thermodynamic - limit approach @xcite as well as to resolve a fine structure of the critical region .
a finite - size scaling technique ( see , for example , the reviews @xcite ) is aimed at this task , but it is phenomenological and mainly oriented to a numerical analysis of the critical scaling .
last , but not least , is an important fact that in the critical region there are no closed dyson - type equations for the constrained , true green s functions . a crucial finding of the present theory is the exact recurrence equations for the partial one- and two - operator contractions .
they follow from a remarkable property of the partial operator contractions to reproduce themselves under a contraction operation .
the previously developed methods did not allow one to account simultaneously for all these complex factors .
we introduce the required additional techniques , including the following novel methods : \(i ) the nonpolynomial diagram technique and partial contraction of operators @xcite , \(ii ) the partial difference ( recurrence ) equations for the operator contractions @xcite ( a discrete analog of the partial differential equations in discrete mathematics @xcite ) , \(iii ) a characteristic function and cumulant analysis @xcite for a joint distribution of the noncommutative observables .
they are required by the very nature of the critical correlations in the constrained systems as well as by the structure of the rigorous many - body theory .
they complement the standard matsubara technique of the green s functions and dyson equation @xcite .
the mean field and related approximations , associated with the landau , bogoliubov , beliaev - popov and similar standard diagrammatic approaches , lead to the incorrect results for the critical behavior and critical exponents .
the renormalization group approach to the constraints in the systems with the critical behavior is also approximate , although it allows one to catch the main power - law critical asymptotics at the wings of the critical region for the models like the non - linear sigma model or the heisenberg model of ferromagnetism . the main point
, however , is that this renormalization group approach is very different from the exact analysis of the theory presented in @xcite and below . in particular
, the latter allows us to formulate the theory of critical phenomena in the spin systems in terms of the constrained spin bosons and treat those constraints exactly . in that way
, we obtain the exact results , like the ones for the 3d ising model ( sect .
5 ) , which yield not only the critical exponents , but a much more full description of the critical phenomena in the entire critical region .
moreover , we can calculate the critical functions and exponents for the 3d systems by obtaining them from those exact solutions , that is , by a regular green - function s method .
it should be compared with the currently used approximate methods , based , for example , on the physically artificial wilson s @xmath0-perturbation of the space dimensionality @xmath1 , real - space renormalization group , high - temperature series , borel s summation , or numerical simulations .
in retrospect , a desire to build a whole theory around just one or two `` main ideas '' , simplifications or approximations , while missing the other similarly important factors of the complex critical phenomenon in phase transitions , was a typical reason for the failure of many previous attempts to solve the problem .
no one of the principle factors could be missed , if we have to derive the correct microscopic theory in the critical region for a given microscopic hamiltonian from the first principles of statistical physics .
an example of a failure in the critical region and a restriction of an analysis to just a well - formed ordered phase is dyson s theory of spin waves in a ferromagnet @xcite .
apparently , due to a lack of the proper mathematical apparatus , first of all , the partial contraction of operators and diagram technique for the nonpolynomial averages , dyson thought that `` the holstein - primakoff formalism is thus essentially nonlinear and unamenable to exact calculations '' . in sect .
4 we refute dyson s conclusion and show how to solve the problem . a well - known `` theory of many - body systems : pair theory '' by arnowitt and girardeau @xcite , that was stimulated by a bardeen - cooper - schrieffer theory of superconductivity , is another example of such a `` one - main - idea '' approximation .
another possible reason is a prejudicial disbelief of many theorists in the very possibility to find a rigorous full solution to such a complicated many - body problem .
the latter is clearly seen , for example , in an attitude of bogoliubov to the problem of finding the microscopic theory of bec and superfluidity @xcite .
a nice exposition on the bogoliubov approach is given in a review @xcite .
in most works on the bec and superfluidity , the rigorous microscopic theory in the critical region is not even addressed due to its complexity ( see @xcite and references therein ) .
for example , hohenberg and martin in their famous review on the microscopic theory @xcite explicitly state that `` we have nothing to say on the difficult problem of the phase transition '' . a recent trend to substitute a consistent microscopic theory of a phase transition in a given system by an approximate calculation of just a restricted subset of its universal features for a phenomenological model , for example , the critical exponents , also contributed to keeping the problem unsolved .
the present paper is dedicated to a great memory of our colleague dick arnowitt and other pioneers of the many - body theory , who paved a way to our understanding of the many - body physics .
the paper is based on the authors talk presented at the symposium in honor of dr .
richard arnowitt , texas a@xmath22 m university , college station , texas , usa , september 19 - 20 , 2014 .
the bec originates from a conservation of the total number of particles @xmath10 in a system : @xmath23 here @xmath24 and @xmath25 are the occupation operators of the nondegenerate ground state @xmath26 and the excited states @xmath27 in a trap .
the creation , @xmath28 , and annihilation , @xmath29 , operators for a @xmath30-eigenstate in a trap obey a canonical commutation relation @xmath31=\delta_{{\bf k},{\bf k'}}$ ] , where @xmath32 is a kronecker s delta .
note that a system of photons in equilibrium inside a black - box cavity does not demonstrate the bec since it lacks a gauge symmetry due to absorption in the cavity s walls and , hence , the number of photons is not its integral of motion .
thus , a rigorous microscopic theory of bec has to be started with an exact reduction of a many - body hilbert space .
it is restricted by the particle - number constraint , which is an integral of motion in virtue of a global gauge symmetry to be broken in the bec phase transition .
let us consider the bec of @xmath10 interacting particles with a mass @xmath33 in a cubic box with volume @xmath34 and periodic boundary conditions .
the trap s one - particle eigenfunctions @xmath35 and energies @xmath36 are specified by a wave - vector quantum number @xmath37 , where @xmath38 is an integer .
the reduction of an unconstrained fock space @xmath39 , which spans all fock states @xmath40 with integer occupations @xmath41 , to a physical , canonical - ensemble subspace @xmath42 is achieved in two steps : @xmath43 first , we restrict the fock space @xmath39 to a fock space @xmath44 with an excluded ground - state component , since its occupation @xmath45 for any fock state is determined by the total excited - states occupation @xmath46 .
second , we limit an allowed number of excitations by the total number of particles loaded in the trap , @xmath47 . according to @xcite , the physics ( dynamics , fluctuations , etc . ) of the system is determined by the creation and annihilation of the canonical - ensemble excitations ( not particles ! ) via the operators @xmath48 which leave invariant ( i.e. , do not lead out of ) the physical , canonical - ensemble subspace @xmath42 .
they are the step - function @xmath49 cutoff of the transition operators @xmath50 which include a proper normalization ; @xmath51 if @xmath52 and @xmath53 if @xmath54 . the operators in eq .
( [ alpha ] ) were introduced in the theory of bec by bogoliubov @xcite in 1947 as well as by girardeau and arnowitt @xcite in 1959 and coincide with the used earlier , in 1940 , by holstein and primakoff @xcite operators for the spin bosons , if a value of a spin is equal to @xmath55 ( see sect .
we introduce also a fock space @xmath56 , which spans only the excited states @xmath57 , and the creation and annihilation operators in that space @xmath56 , @xmath58 and @xmath59 , which obey a canonical commutation relation @xmath60=\delta_{{\bf k},{\bf k'}}$ ] .
their @xmath49-cutoff counterparts are the subspace @xmath61 , which spans only the fock states with a total occupation @xmath62 $ ] not larger than the number of particles @xmath10 , and the operators @xmath63 they constitute a representation , which is exactly isomorphic to the original one in eq .
( [ alpha ] ) and will be used from now on . in the sense of that isomorphism , one has @xmath64 the isomorphism in eq .
( [ alpha=beta ] ) is not trivial and is valid only on @xmath65 .
first , the commutation relations for the creation and annihilation operators of the true excitations in eqs .
( [ alpha ] ) and ( [ hatbeta ] ) are not canonical : @xmath66=\delta_{{\bf k},{\bf k'}}(1-\delta_{\hat{n},n})-\hat{\alpha}_{\bf k'}^{'\dagger } \hat{\alpha}_{\bf k}^ { ' } \delta_{\hat{n},n } , \label{nonccralpha}\ ] ] @xmath67=\delta_{{\bf k},{\bf k'}}(1-\delta_{\hat{n},n})-\hat{\beta}_{\bf k'}^{'\dagger } \hat{\beta}_{\bf k}^ { ' } \delta_{\hat{n},n}. \label{nonccr}\ ] ] second , it does not imply an isomorphism between the operators @xmath68 on @xmath44 and the operators @xmath69 on @xmath56 , because a commutation relation for the latter is canonical , but a commutation relation for the former coincides with the noncanonical one in eq . ( [ nonccralpha ] ) . a free ( zeroth - order ) ,
ideal - gas hamiltonian in the canonical - ensemble hilbert space remains standard : @xmath70 however , the interaction hamiltonian @xmath71 in the physical , constrained hilbert subspace @xmath65 acquires an additional , induced by the constraint , nonlinear interaction determined by the terms with the nonpolynomial operator functions @xmath49 and @xmath72 .
an explicit formula for @xmath71 follows from a standard two - body interaction hamiltonian @xcite @xmath73 the latter is written in the original , unconstrained , auxiliary fock space @xmath39 via a field operator @xmath74 at a position @xmath75 and assuming a symmetric potential @xmath76 .
an exact result is @xcite @xmath77 where we use the true , step - function cutoff field operators @xmath78 @xmath79 for any operator @xmath80 , we introduce a usual matsubara operator @xmath81 , which evolves in the heisenberg representation in accord with a total hamiltonian @xmath82 in an imaginary time @xmath83 $ ] , related to an inverse temperature @xmath84 . a primary question to a microscopic theory is a derivation of the equations for the mean value and correlations of the matsubara annihilation and creation operators @xcite , both the unconstrained ( @xmath85 ) and true ( @xmath86 ) ones . here
@xmath87 and the operators are written in a momentum representation .
below we use also a coordinate representation , where @xmath88 is replaced by @xmath89 .
the definitions of the unconstrained ( auxiliary ) and true coherent order parameters and matsubara green s functions are , respectively , as follows @xmath90 @xmath91 @xmath92 @xmath93 here @xmath94 or @xmath95 mean the averages over the unconstrained or @xmath49-cutoff excited - states hilbert spaces @xmath56 or @xmath96 , respectively . for @xmath49-cutoff operators , like in eqs .
( [ order ] ) and ( [ green ] ) , they differ only by a factor @xmath97 equal to a cumulative probability of a total occupation of the excited states in the space @xmath56 to not exceed the number of particles @xmath10 .
@xmath98 means a standard matsubara @xmath99-ordering .
the indexes @xmath100 and @xmath101 enumerate @xmath102-matrix of the normal and anomalous green s functions , since we set @xmath103 and @xmath104 ( a matsubara - conjugated operator ) for any operator @xmath105 .
both canonical green s functions @xmath106 and @xmath107 in eqs . ( [ green])-([green ] ) are different from the grand - canonical - ensemble ones , employed in a usual beliaev - popov theory @xcite on the unconstrained fock space @xmath39 . the exact hamiltonian in eq . ( [ hint ] )
consists of 7 terms : @xmath108 @xmath109 @xmath110 @xmath111 @xmath112 @xmath113 . in the consistent microscopic theory the heisenberg equation of motion @xmath114
$ ] contains a commutator @xmath115 = ( \tilde{h}_{(n-1)\tau}^{'}-\tilde{h}_{n\tau}^{'})\tilde{\beta}_x + \tilde{a } - \int_v \tilde{a}\frac{d^3 r}{v } , \label{commutator}\ ] ] @xmath116 @xmath117 \frac{d^3 r'}{2},$ ] with an operator @xmath118 , which is incorrectly replaced by a c - number chemical potential @xmath119 in the usual approach .
the hamiltonian @xmath71 turns into the usual beliaev - popov one if one replaces the operators @xmath49 and @xmath72 by c - numbers .
such replacement is wrong near the critical @xmath120-point due to very large critical fluctuations , provided by those operators .
first of all , let us exploit a remarkable fact that , after the first step of the many - body hilbert space reduction in eq .
( [ reduction ] ) and due to the isomorphism in eq .
( [ alpha=beta ] ) , we can work in the unconstrained excited - states hilbert space @xmath56 , where the standard matsubara ( or feynman ) diagram technique and the exact dyson equation are valid .
thus , we immediately get the exact microscopic equations for the unconstrained coherent order parameter and green s functions via a total irreducible self - energy @xmath121 , as it is implied for the dyson - type equations .
the result is @xmath122 , \label{betaeq}\ ] ] @xmath123 .
\label{geq}\ ] ] here the integral operators @xmath124 or @xmath125 , applied to any function @xmath126 , stand for a convolution of that function @xmath126 over the variables @xmath127 with the total irreducible self - energy @xmath128 or a free propagator @xmath129 , respectively : @xmath130\equiv \sum_{j'=1}^2 \int k^{j'x'}_{jx } f_{j'x ' } d^4 x ' \quad \text{for } \
\check{k}=\check{\sigma } , \check{g}^{(0 ) } , \label{keq}\ ] ] where @xmath131 and @xmath132 .
the free propagator , that is the zeroth - order green s function , is determined by a thermal average with the free hamiltonian in eq .
( [ h_0 ] ) , @xmath133 by means of a known for a box , inverse to @xmath125 operator @xcite , one can rewrite eqs .
( [ betaeq ] ) and ( [ geq ] ) in a more familiar differential form : @xmath134 \bar{\beta}_{jx}= -\sum_{j'=1}^2 \int \sigma^{j'x'}_{jx } \bar{\beta}_{j'x ' } d^4 x ' , \label{betaeqd}\ ] ] @xmath135 ( g_{j_1x_1}^{j_2x_2 } + \bar{\beta}_{j_1x_1}\bar{\beta}^{*}_{j_2x_2})= ( -1)^{j_1}\delta_{j_1,j_2}$ ] @xmath136 - \sum_{j'=1}^2 \int \sigma^{j'x'}_{j_1x_1 } g_{j'x'}^{j_2x_2 } d^4 x ' . \label{geqd}\ ] ] all effects of the microscopic interaction between particles on the spontaneous symmetry breaking are encoded in the total irreducible self - energy @xmath137 .
let us stress that already at this step the present microscopic theory of bec is strongly different from the standard beliaev - popov theory .
the reason is that the present self - energy essentially differs from the usually used one in the entire critical region since it is calculated for the full , exact interaction hamiltonian in eq .
( [ hint ] ) and , hence , includes the additional , constraint interaction and all effects of critical fluctuations .
the total irreducible self - energy can be found explicitly from its definition ( @xmath138 ) @xmath139\delta\tilde{\bar{\beta}}_{j_2\kappa_2 } \rangle = ( -1)^{j_1}\sum_{j=1}^2 \int_0^{\frac{1}{t } } \sum_{{\bf k}\neq 0}\sigma_{j_1\kappa_1}^{j\kappa}g_{j\kappa}^{j_2\kappa_2 } d\tau \label{self - energy}\ ] ] with an exact account for a possible coherent ordering @xmath140 .
we substitute @xmath141 in the left - hand - side operator @xmath142 and use the nonpolynomial diagram technique @xcite together with the exact equations for the partial operator contractions , explained in sect .
3.3 . in another way
, we do the first - order , second - order or ladder approximation in interaction , when calculating the left - hand - side average in accord with the main theorem of the diagram technique in an interaction representation for the appropriate free and interaction hamiltonians : @xmath143 of course , a final result does not depend on a way of splitting the total hamiltonian in a particular pair of the free and interaction hamiltonians . both representations @xmath144 in eqs .
( [ h_0 ] ) and ( [ hint ] ) and @xmath145 in eq .
( [ h_0^delta ] ) are equivalent and yield the same total irreducible self - energy ( that is @xmath146 ) , if one takes into account the corresponding equivalent series of diagrams . a function in the left hand side of eq .
( [ geq ] ) is known as an exact unconstrained connected green s function @xmath147 at the equal times @xmath148 , it is defined in accord with an anti - normal ordering of the operators @xmath149 and @xmath150 .
a reason for the latter is explained in sect .
3.3 . in that way ,
the nonpolynomial diagram technique @xcite yields the left hand side in eq .
( [ self - energy ] ) exactly in a form of its right hand side .
the calculations are straightforward , but a full analysis is lengthy and will be presented elsewhere . at its last stage , it is based on a distribution of the noncondensate occupation @xcite , @xmath151 the latter is given by the generating cumulants @xmath152 . in a presence of an arbitrary coherent ordering @xmath153 for a particular case of an ideal gas we find them exactly : @xmath154 , \bar{n}_{\bf k}= ( e^{\frac{\varepsilon_{\bf k}}{t}}-1)^{-1}. \label{rho}\ ] ] the main problem with the microscopic theory for the true , constrained order parameter and green s functions , eqs .
( [ order ] ) and ( [ green ] ) , is that there are no closed equations available for them in the critical region , at least , of the dyson type .
a reason is that they are the averages which include , along with the usual products of the creation and annihilation operators , the nonpolynomial functions , like the ones ( @xmath49 or @xmath72 ) , entering the true excitation operators in eq .
( [ hatbeta ] ) and the exact constraint interaction hamiltonian in eq .
( [ hint ] ) .
the standard diagram methods are not suited for the nonpolynomial averages .
we find a universal solution to that problem by means of the nonpolynomial diagram technique @xcite .
namely , the true order parameter and green s functions in eqs .
( [ order ] ) and ( [ green ] ) are given by the following explicit formulas @xmath155 \rangle /p_n , \quad p_n = \langle \theta(n-\hat{n } ) \rangle , \label{beta'}\ ] ] @xmath156 + \tilde{b}^{j_2}_{j_1}[\theta ] \rangle /p_n .
\label{g'}\ ] ] here the basis partial one- and two - operator contractions , @xmath157 $ ] and @xmath158 $ ] , are the operator - valued functionals , evaluated for the operator functions @xmath159 or @xmath160 .
we use the short notations for the combined indexes @xmath161 and @xmath162 as well as for an index @xmath163 with a complimentary index @xmath164 ; @xmath165 .
an index @xmath166 ( or @xmath167 ) enumerates different times @xmath168 ( or @xmath169 ) in the external operator @xmath170 ( or @xmath171 ) .
the basis partial one - operator contraction @xmath157 $ ] of the matsubara operator @xmath172 with an arbitrary function @xmath173 of the matsubara operators @xmath174 and @xmath175 for the total occupation of the excited states , @xmath176\equiv \mathcal{a}_{\tau_i } t_{\tau}\{\tilde{\beta}_j^{c}f^c(\tilde{n}_{\tau_1},\tilde{n}_{\tau_2 } ) \ } , \label{s}\ ] ] is defined by an explicit formula via the @xmath177 s taylor series : @xmath176= t_{\tau } \sum_{m_1=0}^{\infty}\sum_{m_2=0}^{\infty } \sum_{m'_1=0}^{m_1}\sum_{m'_2=0}^{m_2 } \frac{f^{(m_1,m_2)}}{m_1!m_2!}\ ] ] @xmath178 where @xmath179 .
an examination of the case of two variables @xmath180 is enough for a generic definition .
a symbol @xmath181 denotes an anti - normal ordering , which prescribes a position of the external operator @xmath172 relative to the function @xmath173 and does not affect any other operators positions , which are determined by the @xmath98-ordering . as is explained in the next sect .
3.4 , this definition implies a sum of all possible partial connected contractions , which were used in the nonpolynomial diagram technique in @xcite and denoted by the superscripts `` c '' in eqs .
( [ s ] ) , ( [ b ] ) , ( [ sint ] ) , and ( [ bint ] ) . note that , contrary to the usual notations @xcite , here `` c '' means only the partial ( not full ! ) contraction .
the basis partial two - operator contraction @xmath182 $ ] of the two matsubara operators @xmath183 and @xmath184 with an arbitrary function @xmath173 of the two occupation matsubara operators @xmath174 and @xmath175 , @xmath185 \equiv \mathcal{a}_{\tau_{i_1}\tau_{i_2 } } t_{\tau } \ { \delta \tilde{\beta}_{j_1}^{c}\delta \tilde{\bar{\beta}}_{j_2}^{c } f^c(\tilde{n}_{\tau_1},\tilde{n}_{\tau_2 } ) \ } , \label{b}\ ] ] is defined by an analytical formula similar to eq .
( [ staylor ] ) : @xmath185= t_{\tau } \sum_{m_1=0}^{\infty}\sum_{m_2=0}^{\infty } \sum_{m'_1=0}^{m_1}\sum_{m'_2=0}^{m_2 } \frac{f^{(m_1,m_2)}}{m_1!m_2!}\ ] ] @xmath186 again , the anti - normal ordering @xmath187 prescribes only the positions of the external operators @xmath183 and @xmath184 relative to the function @xmath173 and does not affect any other operators positions , which are determined by the @xmath98-ordering .
the partial operator contractions with the other than anti - normal orderings also exist and could be calculated , but the anti - normally ordered ones in eqs . ( [ s ] ) and ( [ b ] ) are more convenient since the true , constrained creation and annihilation operators in eqs .
( [ alpha ] ) and ( [ hatbeta ] ) are anti - normally ordered relative to their step - functions @xmath188 .
the exact closed difference ( recurrence ) equations for the basis partial operator contractions @xmath189 $ ] and @xmath190 $ ] for an arbitrary function @xmath191 , where a set @xmath192 consists of the integers @xmath193 and @xmath194 , are derived in the subsequent sect .
those equations have the following universal form : @xmath195= \bar{\beta}_j f(\ { m_{j ' } \})+ g_j^{j ' } \delta_{m_{j ' } } \tilde{s}_{j'}[f(\ { m_{j ' } \ } ) ] , \label{1-contraction}\ ] ] @xmath196 = -g_{j_1}^{j_2}f(\ { m_{j ' } \ } ) - g_{j_1}^{j ' } g_{j'}^{j_2 } \delta_{m_{j'}}f(\ { m_{j ' } \})\ ] ] @xmath197 .
\label{2-contraction}\ ] ] a matrix @xmath198 , which enters these equations as the coefficients , is given by the solution of eq .
( [ geq ] ) for the exact unconstrained connected green s function in eq .
( [ g ] ) . for a special case of the equal times
@xmath199 it is defined in accord with an anti - normal ordering of the operators @xmath200 and @xmath201 .
the latter is dictated by the anti - normal ordering in the definition of the partial operator contractions in eqs .
( [ s])-([btaylor ] ) . in eqs .
( [ 1-contraction])-([2-contraction ] ) , a symbol @xmath202 means a partial difference operator @xcite ( @xmath203 and @xmath204 ) and we assume an einstein s summation over the repeated indexes @xmath205 .
the sums run over @xmath206 and all different arguments @xmath207 of the function @xmath177 , enumerated by @xmath208 , @xmath209 , for @xmath210 and similarly for @xmath211 .
the equations ( [ 1-contraction ] ) and ( [ 2-contraction ] ) are the linear systems of the integral equations over the wave vector variables and the discrete ( recurrence ) equations over variables @xmath193 and @xmath194 with the well - known methods of solution @xcite , e.g. , via a z - transform ( a discrete analog of a laplace transform ) , a characteristic function ( a solution for @xmath212 with a subsequent fourier transform ) , or a direct recursion .
the operator contractions in eqs .
( [ beta])-([g ] ) are given by those solutions at @xmath213 .
finally , their average in eqs .
( [ beta])-([g ] ) amounts to the averages like @xmath214 , that is reduced to a calculation of a joint probability distribution @xmath215 of the noncommuting operators @xmath174 and @xmath175 .
the latter can be done similar to @xcite via its characteristic function @xmath216 , which is equal to @xmath217}{z_{\bf k}-z_1z_2}}}{[(z_{\bf k}-z_1z_2)/(z_{\bf k}-1 ) ] } \label{2dchar - function}\ ] ] for any @xmath218 in the zeroth order in interaction ; @xmath219 .
the result in eq .
( [ 2dchar - function ] ) , together with its particular case for @xmath220 in eqs .
( [ rho_n])-([rho ] ) , provides a basis for a perturbation analysis .
the macroscopic wave function ( the coherent order parameter ) of the excitations condensate @xmath221 , entering eqs .
( [ geq ] ) , ( [ geqd ] ) , and ( [ beta])-([2-contraction ] ) , makes the quasiparticles stable and can be found from eq .
( [ betaeq ] ) or eq .
( [ betaeqd ] ) .
we start a derivation of the equations for the operator contractions with a formulation of the nonpolynomial diagram technique for them in the interaction representation with the free hamiltonian @xmath222 in eq .
( [ h_0 ] ) or @xmath223 in eq .
( [ h_0^delta ] ) .
let @xmath224 denotes an interaction representation of any operator @xmath80 , that is @xmath225 or @xmath226 .
first , let us consider an equivalent definition of the basis partial one - operator contraction in the interaction representation . according to the main theorem of the diagram technique @xcite , a thermal average of the matsubara operator @xmath227 over a statistical distribution @xmath228 with a full hamiltonian @xmath229 is equal to a thermal average of the interaction - representation operator @xmath230 over a statistical distribution @xmath231 with a free hamiltonian @xmath222 . here
a symbol @xmath232 stands for a standard matsubara s - matrix @xmath233 we omit a normalization factor @xmath234 , which is canceled since , as usual , we consider only the connected diagrams .
hence , in the interaction representation the basis partial one - operator contraction @xmath235\equiv \mathcal{a}_{\tau_{i_1 } } t_{\tau}\{\hat{\beta}_{j_1}^{c}f^c(\hat{n}_{\tau_{i_1}},\hat{n}_{\tau_{i_2 } } ) s\big(\frac{1}{t}\big ) \ } , \label{sint}\ ] ] corresponding to the matsubara one in eqs .
( [ s])-([staylor ] ) , is equal to @xmath236 plus a sum of all operators obtained from the operator @xmath237 by all possible partial connected contractions via the 1d - connected diagrams , which start from the external operator @xmath238 , connect a series of vertices @xmath239 inside the function @xmath240 via the elementary contractions in eq .
( [ deltag0 ] ) , and end at one of the annihilation or creation operators @xmath241 inside the function @xmath242 .
an elementary contraction is a negative connected free green s function @xmath243 an equivalent definition of the basis partial two - operator contraction in the interaction representation , @xmath244\equiv \mathcal{a}_{\tau_{i_1}\tau_{i_2 } } t_{\tau}\{\delta \hat{\beta}_{j_1}^{c}\delta \hat{\bar{\beta}}_{j_2}^{c } f^c(\hat{n}_{\tau_{i_1}},\hat{n}_{\tau_{i_2 } } ) s\big(\frac{1}{t}\big ) \ } , \label{bint}\ ] ] which corresponds to the matsubara one defined in eqs .
( [ b])-([btaylor ] ) , is similar .
namely , the @xmath245 $ ] is equal to a sum of all operators obtained from the operator @xmath246 by all possible partial connected contractions via the 1d - connected diagrams , which start from one external operator @xmath238 , connect a series of vertices @xmath247 inside the function @xmath240 via the elementary contractions in eq .
( [ deltag0 ] ) , and end at another external operator @xmath248 .
an elementary , generic block of such contractions is a partial contraction of the operator @xmath249 with the @xmath250-th power of the total excited - states occupation operator @xmath251 .
let us make just one elementary contraction ( [ deltag0 ] ) of the external operator @xmath252 with each creation ( @xmath253 ) or annihilation ( @xmath254 ) operators from each factor @xmath255 ( see eqs .
( [ n ] ) and ( [ nj ] ) ) , located at the positions @xmath256 in a sequence of @xmath250 factors in the @xmath257 .
the result of that one - step contraction operation is a following sum : @xmath258 @xmath259,\ ] ] where the operators @xmath260 and @xmath261 depend on a position number @xmath262 , at which the remaining after that one - step contraction operators @xmath263 and @xmath264 are located in the sequence of the @xmath255-factors .
the canonical commutation relations yield @xmath265 and @xmath266 .
the corresponding sums over @xmath262 are reduced to a discrete difference @xmath267 , namely , @xmath268 and @xmath269 \hat{\bar{\beta}}_{{\bf k'}\tau'}$ ] .
then , let us dress the free propagators in eq .
( [ element ] ) into the exact unconstrained connected green s functions ( [ g ] ) by including all appropriate diagrams , coming from the s - matrix in eq .
( [ smatrix ] ) . in this way
, we conclude that the considered one - step elementary contraction is a transformation of the original operator @xmath270 into a convolution of its discrete - difference counterpart @xmath271 with the exact unconstrained connected green s function @xmath198 , that is @xmath272 it follows from the definition in eq .
( [ sint ] ) , that if one supplements this first step by all other required contractions , then the operator @xmath271 in the right hand side of eq .
( [ recursion ] ) will become a basis partial operator contraction @xmath273 $ ] .
therefore , taking into account an obvious term @xmath274 and combining all terms of the taylor series , we immediately arrive to the final eq .
( [ 1-contraction ] ) for the basis partial one - operator contraction . a derivation of the final eq .
( [ 2-contraction ] ) for the basis partial two - operator contraction also immediately follows from a property of the one - step contraction in eq .
( [ recursion ] ) to reproduce an original operator as its difference .
namely , the second term in eq .
( [ 2-contraction ] ) comes from an additional contraction of the operator @xmath275 from the right hand side of eq .
( [ recursion ] ) with the second external operator @xmath248 of @xmath245 $ ] in its definition ( [ bint ] ) .
the third term in eq .
( [ 2-contraction ] ) is a result of a parallel application of the one - step contraction in eq .
( [ recursion ] ) to both external operators @xmath238 and @xmath248 in @xmath245 $ ] . the first term in eq . ( [ 2-contraction ] ) is a result of a direct contraction between the external operators . the derived fundamental equations ( [ betaeq])-([geq ] ) , ( [ beta])-([g ] ) and ( [ 1-contraction])-([2-contraction ] ) consistently account for the critical fluctuations as well as for the effects of the inter - particle interaction on the condensate and quasiparticles .
these equations have a perfectly canonical and proper matrix structure .
the unconstrained order parameter @xmath276 and green s function @xmath277 form the wavevector - dependent 4-component vector and @xmath278-matrix , respectively , since the combined index @xmath279 includes the nambu index @xmath280 and the time @xmath168 index @xmath166 , each with the two values .
the @xmath276 and @xmath277 are the solutions of the canonical dyson - type equations ( [ betaeq])-([geq ] ) or ( [ betaeqd])-([geqd ] ) , which are the integral equations over the wave vector variable , and are determined by the @xmath278-matrices of the free propagator @xmath281 in eq .
( [ g0 ] ) and the total irreducible self - energy @xmath282 , defined by a two - body interaction in eq .
( [ self - energy ] ) . in their turn , these solutions @xmath276 and @xmath277 enter the inhomogeneous linear integral matrix equations ( [ 1-contraction])-([2-contraction ] ) for the basis partial one- and two - operator contractions @xmath283 $ ] ( 4-vector ) and @xmath284 $ ] ( @xmath278-matrix ) via the source terms and the coefficients .
the coefficients are given by the unconstrained connected green s function @xmath278-matrix @xmath285 in eq .
( [ g ] ) ( cf .
a usual nambu s @xmath102-matrix of the normal and anomalous green s functions ) .
in addition to being matrix and integral , the latter equations ( [ 1-contraction])-([2-contraction ] ) are the partial difference equations ( a discrete analog of the partial differential equations ) over the variables @xmath193 and @xmath194 .
the difference equations are required by a discrete , quantum nature of the excitations .
the revealed nontrivial universal structure of these fundamental equations is uniquely prescribed by the microscopic physics of the critical phenomena because these equations are not the model , approximate , or phenomenological ones , but the exact equations .
it is immediate to generalize them to any trapping potentials and boundary conditions .
they open a way to solve the long - standing problem of the bec and other phase transitions @xcite , including a restricted canonical ensemble problem @xcite , and describe numerous modern laboratory and numerical experiments on the critical phenomena in bec of the mesoscopic systems @xcite .
a principle difference of these microscopic equations from the usual gross - pitaevskii and beliaev - popov equations comes from a presence of the operator functions @xmath49 and @xmath72 in the true hamiltonian ( [ hint ] ) .
also , it comes from the symmetry - constrained nature of the actual excitation operators @xmath286 , which determine the true order parameter and green s functions in eqs .
( [ order ] ) and ( [ green ] ) . a replacement of those operator functions by the c - numbers @xmath287 , that is a mean - field approximation , becomes valid only far outside the critical region , at low enough temperatures , when the ordered phase is fully formed and the order parameter is much larger than its fluctuations .
so , the exact microscopic equations asymptotically turn into and , hence , justify the usual gross - pitaevskii and beliaev - popov ones only far outside the critical region .
note that , within the particular case of the homogeneous bec in a box with the periodic boundary conditions , the usually used beliaev - popov equations describe bec as a smooth change of the stable quasiparticles , dressed by a bogoliubov coupling via a condensate occupation and possessing the inter - mode correlations ( the anomalous green s functions ) without any own coherence . in that picture ,
a coherent superfluid flow is generated by some external sources or boundary motion , but an appearance of the coherent macroscopic order parameter @xmath288 is excluded by the hugenholtz - pines ( @xmath289 ) or hohenberg - martin ( @xmath290 ) condition @xcite @xmath291 , that ensures a stable gapless spectrum of quasiparticles .
that usual mechanism of bec is similar to the one in an ideal gas , being just its dressed - by - interaction version .
it differs from the outlined general mechanism , allowing an instability of restructuring of the condensate by @xmath292 , that is a symmetry breaking ( cf . a density wave in a superfluid @xcite ) .
a detailed analysis of eqs . ( [ betaeq])-([geq ] ) , ( [ beta])-([g ] ) and ( [ 1-contraction])-([2-contraction ] ) will be given elsewhere .
the major point is that the microscopic theory is valid both inside and outside the critical region . in particular
, it is capable of a microscopic calculation of the critical exponents and functions as well as other parameters of the bec phase transition near the @xmath120-point for an actual , canonical - ensemble mesoscopic system , as opposed to a grand - canonical - ensemble bulk - limit model , usually studied in the phenomenological renormalization - group approach @xcite .
there is another known limit of this microscopic theory .
namely , for a vanishing interaction , when @xmath293 and @xmath294 in eqs .
( [ rho ] ) and ( [ 2dchar - function ] ) , it coincides with a solution for the critical fluctuations in the bec of a mesoscopic ideal gas @xcite , which is its zeroth - order approximation . in short , all these advances come from the correct account , first , of the noether s symmetry constraints of the many - body hilbert space and , second , of the related properties of the true excitations in a mesoscopic system .
the microscopic theory of phase transitions , described above , is truly universal .
let us demonstrate that fact by formulating the microscopic theory of the magnetic phase transitions in a form , which is completely similar to the bec one , developed in sect .
the only important difference is that in the case of the magnetic phase transition there are many local constraints ( [ spin ] ) for each of n spins in a lattice , instead of just one global particle - number constraint ( [ n ] ) in the case of bec in a gas .
let us consider a cubic lattice of @xmath10 interacting immovable spins in a box with a volume @xmath34 and the periodic boundary conditions .
a magnitude of each spin is the same @xmath295 .
the lattice sites of the spins are enumerated by a position vector @xmath75 . according to a well - known holstein - primakoff representation @xcite , worked out also by schwinger @xcite ( see @xcite ) , each spin is a system of the two spin bosons , which are the harmonic oscillators constrained to have a fixed total occupation @xmath296 the creation ( @xmath297 , @xmath298 ) and annihilation ( @xmath299 , @xmath300 ) operators obey the bose canonical commutation relations : @xmath301=\delta_{{\bf r},{\bf r'}}$ ] , @xmath302=\delta_{{\bf r},{\bf r'}}$ ] , and all @xmath303-operators commute with all @xmath304-operators .
a vector spin operator @xmath305 at a site @xmath75 is given by its components as follows @xmath306 the spin raising and lowering operators are equal to @xmath307 similar to sect .
( [ reduction ] ) , we make the two - step reduction of an unconstrained fock space @xmath308 , which spans all fock states @xmath309 of the two spin bosons at a site @xmath75 , to a physical subspace @xmath310 as follows @xmath311 first , we restrict the fock space @xmath308 to a fock space @xmath312 by excluding the @xmath313-boson component , since its occupation @xmath314 for any fock state is determined by the @xmath303-boson s occupation @xmath7 .
second , we limit an allowed number of excitations of the @xmath303-boson in the physical hilbert space @xmath310 by an interval @xmath315 .
the physics ( dynamics , fluctuations , etc . ) of each spin is determined by the creation and annihilation of the canonical - ensemble excitations ( not bosons themselves ! ) in the system of two spin bosons via the operators @xmath316 which leave invariant ( i.e. , do not lead out of ) the physical subspace @xmath310 .
they are the step - function @xmath317 cutoff of the transition operators @xmath318 which are exactly similar to the bogoliubov - girardeau - arnowitt ones in eq .
( [ alpha ] ) .
for every site @xmath75 , we introduce also a fock space @xmath319 , which spans the states @xmath320 of an unconstrained excitation at the site @xmath75 , and the creation and annihilation operators @xmath321 and @xmath322 in that space , which obey a canonical commutation relation @xmath323=\delta_{{\bf r},{\bf r'}}$ ] .
their @xmath317-cutoff counterparts are the subspace @xmath324 , which spans only the fock states with a total occupation @xmath325 $ ] , and the operators @xmath326 they constitute a representation , which is exactly isomorphic to the original one in eq .
( [ spinalpha ] ) .
it is valid in the total physical many - body hilbert space of the system of @xmath10 spins @xmath327 , which is a tensor product of all physical hilbert subspaces @xmath324 , and will be used from now on . in the sense of that isomorphism
, one has @xmath328 the isomorphism in eq .
( [ spinalpha=beta ] ) is not trivial and is valid only on @xmath329 .
first , the commutation relations for the creation and annihilation operators of the true spin excitations in eqs .
( [ spinalpha ] ) and ( [ spinhatbeta ] ) are not canonical : @xmath330=\delta_{{\bf r},{\bf r'}}(1- ( 2s+1 ) \delta_{\hat{n}_{\bf r},2s } ) , \label{spinnonccralpha}\ ] ] @xmath331=\delta_{{\bf r},{\bf r'}}(1- ( 2s+1 ) \delta_{\hat{n}_{\bf r},2s } ) .
\label{spinnonccr}\ ] ] second , it does not imply an isomorphism between the operators @xmath332 on @xmath312 and the operators @xmath333 on @xmath319 , because the commutation relations for the latter are canonical , but the commutation relations for the former coincide with the noncanonical ones in eq .
( [ spinnonccralpha ] ) . finally , we present the explicit formulas for the vector components of the spin operator in eqs . ( [ spinvector])-([spin+- ] ) in terms of the creation and annihilation operators of the true spin excitations in eq . ( [ spinhatbeta ] ) : @xmath334 @xmath335 a free hamiltonian of a system of @xmath10 spins in a lattice @xmath336 is determined by a zeeman energy @xmath337 of a spin in an external magnetic field @xmath6 ( which is assumed homogeneous and directed along the axis @xmath338 ) via a @xmath339-factor and a bohr magneton @xmath340 .
we intentionally define the free hamiltonian in eq .
( [ spinh_0 ] ) via the unconstrained occupation operators @xmath341 on the extended many - body hilbert space @xmath342 without any @xmath317 cutoff factors .
that makes the free hamiltonian purely quadratic which is necessary for a validity of the standard diagram technique .
the latter is crucial for a derivation of the dyson - type equations , like eqs .
( [ spinbetaeq ] ) and ( [ spingeq ] ) .
one is allowed to skip the @xmath317 cutoff factors in @xmath222 because of ( i ) an equality @xmath343 , which is valid on the physical many - body hilbert space @xmath329 , and ( ii ) a fact that the occupation operator @xmath341 leaves invariant that space @xmath329 . however , the interaction hamiltonian @xmath344 in the physical , constrained many - body hilbert space @xmath329 includes all terms with the nonpolynomial operator functions : the @xmath317-cutoff factors and the functions @xmath345 , coming from the @xmath346 and @xmath347 components of spins . as a generic example
, we consider a well - known heisenberg hamiltonian @xcite @xmath348 where a coupling between spins is usually assumed to be a symmetric function @xmath349 of a vector @xmath350 , connecting spins .
it contains the true spin excitations , eq .
( [ spinhatbeta ] ) , and , hence , should be written via the unconstrained excitation operators @xmath351 and @xmath322 with an explicit account for all @xmath317-cutoff factors and functions @xmath345 , because one needs an exact account for all spin - constraint nonlinear interactions in the critical region .
thus , we find an exact result for the interaction hamiltonian , presented here as a sum of two terms : @xmath352 @xmath353 the first one is related to the xy model , @xmath354 @xmath355 the second one is related to the ising model , @xmath356 [ s-\theta ( 2s-\hat{n}_{\bf r'})\hat{n}_{\bf r ' } ] . \label{hising}\ ] ] the result in eqs . ( [ hxyhz])-([hising ] ) agrees with the holstein - primakoff s one @xcite , but includes the additional @xmath317-cutoff factors , which are very important for the rigorous microscopic theory of magnetic phase transitions in the critical region . a total hamiltonian @xmath357 makes any operator @xmath80 , evolving in an imaginary time @xmath83 $ ] in the heisenberg representation , the matsubara operator @xmath81 .
let us introduce some notations .
a symbol @xmath358 stands for an operator itself @xmath359 at @xmath360 and for a matsubara - conjugated operator @xmath361 at @xmath362 .
a four - dimensional coordinate is @xmath89 .
a product of all cutoff factors in a lattice of @xmath10 spins is @xmath363 the quantities of primary interest in a microscopic theory of magnetic phase transitions are the mean value and correlations of the matsubara annihilation and creation operators for spin excitations , both the unconstrained ( @xmath364 ) and true ( @xmath365 ) ones .
the definitions of the unconstrained ( auxiliary ) and true coherent order parameters and matsubara green s functions for spin excitations are as follows @xmath366 @xmath367 @xmath368 @xmath369 here @xmath94 or @xmath370 mean the averages over the unconstrained or @xmath371-cutoff many - body hilbert spaces @xmath56 or @xmath372 , respectively .
they differ by the @xmath371-cutoff factor from eq .
( [ spintheta ] ) and a normalization factor @xmath373 , which is equal to a cumulative probability of all occupations of the spin excitations in the space @xmath56 to be within the physically allowed intervals : @xmath374 $ ] for all lattice sites @xmath75 . as it was done in sect .
3.2 , we again use a remarkable fact that , after the first step of the many - body hilbert space reduction in eq .
( [ spinreduction ] ) and due to the isomorphism in eq .
( [ spinalpha=beta ] ) , it is possible to work in the unconstrained many - body hilbert space @xmath56 , where the standard matsubara ( or feynman ) diagram technique and the exact dyson equation are valid .
thus , we immediately get the exact microscopic equations for the unconstrained coherent order parameter and green s functions of spin excitations via a total irreducible self - energy @xmath121 as follows @xmath122 , \label{spinbetaeq}\ ] ] @xmath123 .
\label{spingeq}\ ] ] here the integral operators @xmath124 or @xmath125 , applied to any function @xmath126 , stand for a convolution of that function @xmath126 over the variables @xmath127 with the total irreducible self - energy @xmath128 or a free propagator @xmath129 , respectively : @xmath130\equiv \sum_{j'=1}^2 \sum_{\bf r ' } \int_0^{1/t } k^{j'x'}_{jx } f_{j'x ' } d \tau ' \quad \text{for } \
\check{k}=\check{\sigma } , \check{g}^{(0)}. \label{spinkeq}\ ] ] the free propagator is determined by exactly the same eq .
( [ g0 ] ) , that was used in sect .
3.2 , but with an appropriate for spins free hamiltonian in eq .
( [ spinh_0 ] ) .
the total irreducible self - energy is defined by equation @xmath375\delta\tilde{\bar{\beta}}_{j_2x_2 } \rangle = ( -1)^{j_1}\sum_{j=1}^2 \int_0^{\frac{1}{t } } \sum_{\bf r}\sigma_{j_1x_1}^{jx}g_{jx}^{j_2x_2 } d\tau \label{spinself - energy}\ ] ] and exactly accounts for a possible coherent ordering @xmath376 .
we can substitute @xmath377 in the left - hand - side operator @xmath378 and use the nonpolynomial diagram technique @xcite together with the exact equations for the partial operator contractions , explained in sections 3.3 and 4.3 , or do the first - order , second - order or ladder approximation in interaction , when calculating the left - hand - side average in accord with the main theorem of the diagram technique in the interaction representation for the appropriate free and interaction hamiltonians : @xmath379 of course , a final result does not depend on a way of splitting the total hamiltonian in a particular pair of the free and interaction hamiltonians . both representations @xmath380 in eqs .
( [ spinh_0])-([spinhint ] ) and @xmath381 in eq .
( [ spinh_0^delta ] ) are equivalent and yield the same total irreducible self - energy ( that is @xmath146 ) , if one takes into account the corresponding equivalent series of diagrams . a function in the left hand side of eq .
( [ spingeq ] ) is known as an exact unconstrained connected green s function @xmath382 , defined in eq .
( [ g ] ) .
a strategy for calculation of the self - energy is similar to the bec one ( sect .
its full analysis is lengthy and will be discussed elsewhere .
the true , constrained order parameter and green s functions of spin excitations , eqs .
( [ spinorder ] ) and ( [ spingreen ] ) , do not obey any standard closed equations of the dyson type .
a reason is that they are the averages which include , along with the usual products of the creation and annihilation operators , the nonpolynomial functions , like the ones ( @xmath317 or @xmath383 ) entering the true spin - excitation operators in eq .
( [ spinhatbeta ] ) and the exact interaction hamiltonian in eq .
( [ spinhint ] ) .
a standard diagram technique is not suited to deal with the nonpolynomial averages . as in sect .
3.3 , we solve that problem by means of the nonpolynomial diagram technique @xcite .
namely , the true order parameter and green s functions in eqs .
( [ spinorder ] ) and ( [ spingreen ] ) are given by the following explicit formulas @xmath384 \rangle /p_s , \quad p_s = \langle \hat{\theta } \rangle , \label{spinbeta'}\ ] ] @xmath385 + \tilde{b}^{j_2}_{j_1}[\tilde{\theta}_{\tau_1 } \tilde{\theta}_{\tau_2 } ] \rangle /p_s .
\label{sping'}\ ] ] here the basis partial one- and two - operator contractions , @xmath157 $ ] and @xmath158 $ ] , are the operator - valued functionals , evaluated for an operator function @xmath177 .
we continue to use the short notations for the combined indexes , but this time in the coordinate , rather than momentum , representation that is more convenient for a lattice .
those indexes are @xmath386 and @xmath387 as well as @xmath388 with a complimentary index @xmath164 ; @xmath165 .
an index @xmath166 ( or @xmath167 ) enumerates different times @xmath168 ( or @xmath169 ) in the external operator @xmath389 ( or @xmath390 ) .
a definition of the basis partial operator contractions is similar to the one discussed in sections 3.3 and 3.4 .
let us consider a generic case of an arbitrary operator function @xmath391 , which depends on the two sets @xmath392 and @xmath393 of the spin - excitation occupation operators at all lattice sites at two different times @xmath394 and @xmath395 .
the basis partial one - operator contraction @xmath396\equiv \mathcal{a}_{\tau_i } t_{\tau}\{\tilde{\beta}_j^{c}f^c(\{\tilde{n}_{x_1},\tilde{n}_{x_2 } \ } ) \ } \label{spins}\ ] ] is defined by an explicit formula via the @xmath177 s taylor series : @xmath397= t_{\tau } \sum_{\{m_{x_1}\}}^{\infty}\sum_{\{m_{x_2}\}}^{\infty } \sum_{\{m'_{x_1}\}}^{\{m_{x_1}\}}\sum_{\{m'_{x_1}\}}^{\{m_{x_2}\ } } \frac{f^{(\{m_{x_1}\},\{m_{x_2}\})}}{\prod_{\{x_1,x_2\ } } ( m_{x_1}!m_{x_2}!)}\ ] ] @xmath398 @xmath399 here the sums run over all nonnegative integers @xmath400 and @xmath401 . a symbol @xmath181 denotes an anti - normal ordering , which prescribes a position of the external operator @xmath172 relative to the function @xmath402 and does not affect any other operators positions , which are determined by the @xmath98-ordering . as is explained in sect .
3.4 , this definition implies a sum of all possible partial connected contractions , which were used in the nonpolynomial diagram technique in @xcite and denoted by the superscripts `` c '' in eqs .
( [ spins ] ) , ( [ spinb ] ) . in the case ,
when the function @xmath177 depends only on a one - time set of spin - excitation occupations @xmath403 , the formulas for the basis partial operator contractions become much simpler .
the basis partial two - operator contraction @xmath404 \equiv \mathcal{a}_{\tau_{i_1}\tau_{i_2 } } t_{\tau } \ { \delta \tilde{\beta}_{j_1}^{c}\delta \tilde{\bar{\beta}}_{j_2}^{c } f^c(\{\tilde{n}_{\tau_1},\tilde{n}_{\tau_2}\ } ) \ } \label{spinb}\ ] ] is defined by a similar analytical formula @xmath405= t_{\tau } \sum_{\{m_{x_1}\}}^{\infty}\sum_{\{m_{x_2}\}}^{\infty } \sum_{\{m'_{x_1}\}}^{\{m_{x_1}\}}\sum_{\{m'_{x_1}\}}^{\{m_{x_2}\ } } \frac{f^{(\{m_{x_1}\},\{m_{x_2}\})}}{\prod_{\{x_1,x_2\ } } ( m_{x_1}!m_{x_2}!)}\ ] ] @xmath406 an anti - normal ordering @xmath187 prescribes only positions of the external operators @xmath183 and @xmath184 relative to the function @xmath402 and does not affect any other operators positions , which are determined by the @xmath98-ordering .
the exact closed difference ( recurrence ) equations for the basis partial operator contractions @xmath189 $ ] and @xmath190 $ ] for an arbitrary function @xmath407 , where a set @xmath408 consists of the two sets of integers @xmath409 and @xmath410 , can be derived along the same lines as it was done for bec in sect .
those equations have exactly the same universal form as eqs .
( [ 1-contraction ] ) and ( [ 2-contraction ] ) : @xmath195= \bar{\beta}_j f(\ { m_{j ' } \})+ g_j^{j ' } \delta_{m_{j ' } } \tilde{s}_{j'}[f(\ { m_{j ' } \ } ) ] , \label{spin1-contraction}\ ] ] @xmath196 = -g_{j_1}^{j_2}f(\ { m_{j ' } \ } ) - g_{j_1}^{j ' } g_{j'}^{j_2 } \delta_{m_{j'}}f(\ { m_{j ' } \})\ ] ] @xmath197 .
\label{spin2-contraction}\ ] ] a matrix @xmath198 , which enters these equations as the coefficients , is given by the solution of eq .
( [ spingeq ] ) for the exact unconstrained connected green s function in eq .
( [ g ] ) . for a special case of the equal times @xmath199 , it is defined in accord with an anti - normal ordering of operators
@xmath200 , @xmath201 .
the latter is dictated by the anti - normal ordering in the definition of partial operator contractions in eqs .
( [ spins])-([spinbtaylor ] ) . in eqs .
( [ spin1-contraction])-([spin2-contraction ] ) , a symbol @xmath202 means a partial difference operator @xcite , defined after eq .
( [ 2-contraction ] ) , and we assume an einstein s summation over the repeated indexes @xmath205 .
the sums run over @xmath206 and all different arguments @xmath411 of the function @xmath177 , enumerated by @xmath208 and @xmath412 , for @xmath210 and similarly for @xmath211 .
the equations ( [ spin1-contraction ] ) and ( [ spin2-contraction ] ) are the linear systems of the integral equations over the spin positions variables and the discrete ( recurrence ) equations over the variables @xmath413 with the well - known methods of solution @xcite , e.g. , via a z - transform ( a discrete analog of a laplace transform ) , a characteristic function ( a solution for @xmath414 with a subsequent fourier transform ) , or a direct recursion .
the operator contractions in eqs .
( [ spinbeta])-([sping ] ) are given by those solutions at @xmath415 . finally , their average in eqs .
( [ spinbeta])-([sping ] ) amounts to the averages like @xmath416 , that is reduced to a calculation of a joint probability distribution @xmath417 of the noncommuting operators @xmath418 and @xmath419 .
the latter can be done similar to @xcite by the method of the characteristic function , mentioned in sect .
3.3 ( see an example in sect . 5.2 , 5.4 ) .
the macroscopic wave function ( the coherent order parameter ) of the spin excitations coherence @xmath221 , entering eqs .
( [ spingeq ] ) , ( [ spinbeta ] ) and ( [ spin1-contraction ] ) , makes the quasiparticles stable and can be found from eq .
( [ spinbetaeq ] ) .
the exact results for the constrained green s functions , the order parameter and the partial operator contractions of the presented above microscopic theory provide the regular , canonical tools for finding the exact solutions for the particular many - body hamiltonians , including the 3d systems .
this is in contrast to the known exactly solvable models , the solutions for which are based on some special mathematical tricks , applicable only due to very specific degenerate properties of those models and their low ( 1d or 2d ) dimensionality ( see , for example , @xcite ) . typically , those models , for example , the 2d ising models , demonstrate a critical behavior , which is very different from the behavior of the actual 3d systems . as a generic example , we discuss here the exact solution for the 3d ising model , finding of which is a long - standing problem .
let us consider @xmath10 spins @xmath420 in a lattice , described in sect .
4.1 , with the free hamiltonian @xmath222 in eq .
( [ spinh_0 ] ) and the interaction hamiltonian @xmath421 in eq .
( [ hising ] ) in the absence of the term @xmath422 in eq .
( [ hxyhz ] ) and zero coherent ordering @xmath423 . for a spin at a site @xmath424
there is the coordination number @xmath425 of the neighboring spins at sites @xmath426 which have the nonzero couplings @xmath427 .
the coupling @xmath428 between the spin @xmath424 and any other spin @xmath429 vanishes .
we assume an arbitrary dimensionality of the lattice @xmath430 .
a quantity of primary interest is the equal - time anti - normally ordered correlation matrix @xmath431 , which is equal to the unconstrained connected green s function @xmath432 at @xmath433 .
from now on we use the combined indexes @xmath434 , etc .
the first , main step is finding an exact explicit formula for the total irreducible self - energy from eq .
( [ spinself - energy ] ) : @xmath435 ( g^{-1})_{i'}^{i } .
\label{isingselfenergy}\ ] ] here @xmath436=\langle\tilde{b}_{i_0}^{i'}[f]\rangle$ ] is an unconstrained average of the basis partial two - operator contraction in eq .
( [ spinb ] ) , @xmath437 and @xmath438 denotes the matrix , which is inverse to @xmath339 .
the exact recurrence eq .
( [ spin2-contraction ] ) immediately proves that for a given spin @xmath424 the self - energy matrix in eq .
( [ isingselfenergy ] ) is not zero only for @xmath439 . in other words
, it has a pure @xmath440-banded diagonal structure in indexes @xmath441 : @xmath442 moreover , we can easily and exactly find each @xmath443-matrix block @xmath444 by solving the recurrence eq .
( [ spin2-contraction ] ) .
that result is crucial for the exact solution of the ising model . for simplicity s sake
, we consider below only a particular , usually discussed case of the homogeneous phases , when all sites have the same auto - correlation matrices and other properties . in this case
the order parameter is homogeneous along the lattice and the green s function @xmath445 does not depend separately on each of the two position vectors @xmath446 and @xmath447 , but depends only on their difference @xmath448 .
so , it is a toeplitz matrix with respect to the indexes @xmath446 and @xmath447 .
a generalization to an arbitrary case will be presented elsewhere .
the corresponding @xmath443-matrices @xmath449 $ ] over the indexes @xmath450 and @xmath451 are found as follows @xmath452 @xmath453 here , for a given pair of neighboring sites @xmath454 , we work with the correlation hermitian @xmath455-matrix @xmath456 , @xmath457 where the lattice - site indexes @xmath458 and @xmath459 run only over two values @xmath460 and @xmath461 means the anti - normal ordering .
we introduce the correlation @xmath443-matrix @xmath462 as well as the shortened notations for its auto - correlation , @xmath463 , and cross - correlation , @xmath464 , versions : @xmath465 deriving eq .
( [ bl ] ) , we introduce the @xmath443-matrices @xmath466 needed for the inversion of the block matrix @xmath467 in eq .
( [ q4x4 ] ) , @xmath468 by means of the well - known frobenius formulas .
the result in eq .
( [ bl ] ) includes the single - site and two - sites , joined uncutoff probability distributions , @xmath469 for the spin - boson occupation operators @xmath470 and @xmath471 at the neighboring sites @xmath424 and @xmath472 to acquire the @xmath473 and @xmath474 quanta of excitations , respectively . in sect .
5.2 we calculate those occupation distributions exactly as well .
the self - energy blocks @xmath475 in eq .
( [ p+1selfenergyising ] ) for neighboring sites @xmath476 , can be calculated via a product of two @xmath455-matrices , in which the index @xmath459 in @xmath477 runs only over the two sites @xmath478 , as follows @xmath479 ( q^{-1}(l))_{j'{\bf r'}}^{j{\bf r_0 } } , \label{sigma4x4}\ ] ] @xmath480 ( q^{-1}(l))_{j'{\bf r'}}^{j{\bf r_l}}.\ ] ] a result for the @xmath443-matrix blocks @xmath481 is @xmath482 @xmath483 , \label{sigma(0)}\ ] ] @xmath484 @xmath485(1+c^{\dagger}kcq)\}. \label{sigma(l)}\ ] ] the second step is finding the distributions in eq .
( [ rho01 ] ) .
the single - site distribution of the spin - boson excitations @xmath486 \rangle}{n+1 } , \
n=0,1,2 , \dots , \label{rho - b}\ ] ] can be calculated via the partial two - operator contraction @xmath487 $ ] again by means of the recurrence eq .
( [ spin2-contraction ] ) , which yields the following recurrence equation for @xmath488 : @xmath489\rho_n = 0 ; \label{rho - recurrence}\ ] ] @xmath490 we find its solution via the well - known jacobi polynomial @xmath491 or hypergeometric function , defined by a generalized hypergeometric series @xmath492 : @xmath493 @xmath494^{\frac{1}{2 } } , \
\zeta_{1,2 } = \frac{g_{1{\bf r}}^{1{\bf r}}\mp |g_{1{\bf r}}^{2{\bf r}}|}{1+g_{1{\bf r}}^{1{\bf r } } \mp |g_{1{\bf r}}^{2{\bf r}}| } .
\label{rho0}\ ] ] another way to find the same @xmath488 is to use its relation @xmath495 to a characteristic function @xmath496 , obeying an equation @xmath497 \rangle , \quad \theta(u ) \equiv
\langle e^{iu\tilde{n}_{\bf r } } \rangle .
\label{theta}\ ] ] in case of a function @xmath498 , the recurrence eq .
( [ spin2-contraction ] ) for the @xmath443-matrix of the two - operator contraction @xmath499)$ ] becomes a simple @xmath443-matrix equation @xmath500 we find its solution , plug it into eq .
( [ theta ] ) , and solve that equation .
a result for the characteristic function is @xmath501^{\frac{1}{2 } } \equiv \frac{\rho_0 \sqrt{\zeta_1\zeta_2}}{\sqrt{(z-\zeta_1)(z-\zeta_2 ) } } .
\label{thetafinal}\ ] ] plugging it into eq .
( [ rhotheta ] ) , we obtain the same distribution ( [ rhojacobi ] ) that was found above by a completely different method of eq .
( [ rho - b ] ) .
in particular , we find the probability @xmath502 requested by eqs .
( [ sigma(0 ) ] ) and ( [ sigma(l ) ] ) for the self - energy .
thus , the problem of finding the single - site probability distribution of the spin boson excitations via the auto - correlation coefficients @xmath503 and @xmath504 is fully solved .
a solution of a similar problem for the joined distribution of spin excitations at two sites @xmath505 in eq .
( [ rho01 ] ) , @xmath506 via the two equations for the characteristic function @xmath507 [ theta2 ] @xmath508 \rangle , k=1,2 , \label{theta2}\ ] ] can be done in the same way . in that case of a function @xmath509 , the recurrence eq .
( [ spin2-contraction ] ) for the @xmath455-matrix of the two - operator contraction @xmath510)$ ] , where @xmath511 and @xmath512 , becomes a more complex @xmath455-matrix equation @xmath513 here the @xmath455-matrix @xmath467 is defined in eq .
( [ q - block ] ) by the @xmath443-matrices @xmath514 and @xmath515 of the auto- and cross - correlations for two sites @xmath505 . a matrix eq.([b4x4 ] ) has a well - known in algebra type and can be solved by the regular methods .
moreover , in our case we find a simple explicit solution @xmath516 plugging it into eqs .
( [ theta2 ] ) and solving them , we find an explicit exact solution for the characteristic function @xmath517 where we use the shortened notations for the basis normal and anomalous auto- and cross - correlations @xmath518 and a quadratic , in each variable @xmath519 , polynomial , defined by a determinant of a following @xmath455-matrix , @xmath520 this result yields the joined probabilities of two neighboring sites occupations @xmath215 , entering the self - energy in eqs .
( [ sigma(0)])-([sigma(l ) ] ) , as the simple derivatives in eq . ( [ rhotheta2 ] ) of an elementary function in eq .
( [ theta2final ] ) . for simplicity s sake
, we present them for the case , when @xmath521 and the cross - correlation between neighboring sites is symmetric : @xmath522 .
the former can be always achieved by adjusting phases of the spin boson creation operators .
the latter implies a symmetry : @xmath523[(g_{11}+\xi_2)^2 -g_{12}^2]-2(g_{11}+\xi_1)\times\ ] ] @xmath524 @xmath525 then , a direct differentiation yields the probabilities @xmath526 @xmath527 , \label{rho10}\ ] ] @xmath528 [ 1 + ( q^{-1})_{1{\bf r_2}}^{1{\bf r_2 } } ] \label{rho11}\ ] ] @xmath529 thus , we find the two - sites , joined probability distribution of the spin boson excitations via the basis normal and anomalous auto- and cross - correlations in eq .
( [ g12c12 ] ) .
so , the total irreducible self - energy and the spin - boson occupation statistics are known exactly via @xmath530 basis correlation @xmath443-matrices @xmath531 , in eq .
( [ sandc ] ) alone .
actually , as is shown below , the green s functions , the order parameter and other statistical and thermodynamic quantities are also determined by those auto - correlations @xmath532 for a spin boson at a site @xmath424 and its cross - correlations @xmath533 with the coordination number @xmath425 of the neighboring spins at sites @xmath534 .
due to the complex - conjugation relations @xmath535 there are only two independent , normal @xmath536 and anomalous @xmath537 , correlation parameters per each basis correlation @xmath443-matrix , that is , only @xmath538 numbers , which determine the characteristics of critical phenomena . therefore , the very crucial , third step is to find the exact consistency equations for those @xmath538 basis auto- and cross - correlations .
we solve this problem in two steps .
first , we solve the dyson - type eq .
( [ spingeq ] ) for the unconstrained green s functions in terms of those basis correlations .
second , we close the loop by expressing the basis correlations themselves via those green s functions . for the considered stationary homogeneous phases , the green s functions , the equal - time correlation functions , and the self - energy depend only on the differences of their arguments @xmath539 and @xmath540 , that is @xmath541 hence , it is straightforward to solve the dyson - type eq .
( [ spingeq ] ) by means of the fourier transformation over imaginary time @xmath542 $ ] and the discrete fourier transformation over space .
the latter has a following form @xmath543 where the sums run over all lattice sites @xmath75 with a period @xmath18 and discrete wave vectors @xmath544 with an integer @xmath467 ; @xmath545 $ ] .
we discern the fourier transform and its inverse by the arguments @xmath30 and @xmath75 .
the result for the normal and anomalous green s functions is @xmath546e^{i\omega^{(j)}(\frac{\text{sign}(\tau)}{2t}-\tau)}}{2(\omega^{(2)}-\omega^{(1)})\sin [ \omega^{(j)}/(2 t ) ] } , \label{g11}\ ] ] @xmath547 } , \label{g12}\ ] ] where the two quasiparticle eigen - energies @xmath548^{\frac{1}{2 } } \label{omega12}\ ] ] depend on the wave vector @xmath30 via the self - energies @xmath549 the @xmath443-matrix blocks @xmath550 are found in eqs .
( [ sigma4x4])-([sigma(l ) ] ) .
the spatial fourier transforms of the normal and anomalous equal - time correlation functions follow from eqs .
( [ g11 ] ) and ( [ g12 ] ) in the limit @xmath551 : @xmath552}{i(\omega^{(1)}-\omega^{(2)})[1-\exp ( -i\omega^{(j)}/t ) ] } , \label{g11}\ ] ] @xmath553 } .
\label{g12}\ ] ] thus , we derive the equations for the values of normal and anomalous correlation functions at @xmath530 difference position vectors @xmath554 of the neighboring spins : @xmath555 their right hand side is determined by a left hand side @xmath556 itself via eqs .
( [ sigma4x4])-([sigma(l ) ] ) , ( [ omega12])-([g12 ] ) .
they constitute an exact system of the @xmath538 consistency equations .
its finding means a solution to the ising problem in the same sense as finding of a self - consistency equation in the mean - field theory means a solution to a phase transition problem .
it is straightforward to analyze these explicit consistency equations by the well - developed in the mean - field theory analytical and numerical methods .
this result provides a regular way for the exact calculation of the critical exponents and the critical functions for the 3d systems . in the ising model
the order parameter is equal to the spontaneous magnetization of a spin at a lattice site , @xmath557 for the spin value @xmath420 , it is reduced to a quantity @xmath558 determined by the @xmath371-cutoff , true probability @xmath559 of the spin boson at the site @xmath75 to have one quantum of excitation .
note that the coherent order parameter in eqs .
( [ spinorder ] ) and ( [ spinorder ] ) is absent in the ising model .
the definition of the true , constrained green s function in eq .
( [ spingreen ] ) in the ising model can be rewritten in the form of eq .
( [ sping ] ) , @xmath560 \rangle /p_s$ ] , via the partial two - operator contraction , taken for the function @xmath561 , which is determined by the cutoff factor in eq .
( [ spintheta ] ) .
we find an exact solution of the recurrence eq .
( [ spin2-contraction ] ) for that operator contraction at equal times : @xmath562f .
\label{b(theta)ising}\ ] ] using this result , we find the true , constrained correlation functions @xmath563 of the spin bosons in a lattice : @xmath564 where @xmath565 they are determined by the matrix @xmath566 , which is inverse to the matrix @xmath567 of the unconstrained correlation functions , calculated in the previous sect .
the inverse matrix @xmath438 can be calculated by a technique of the toeplitz matrices , known from the theory of the 2d ising model . a detailed analysis of the obtained results for the true correlation functions and order parameter as well as the true single - site and two - sites , joined probability distributions @xmath568 and @xmath569 of the spin boson occupations , which are the @xmath371-cutoff versions of the calculated in sect .
5.2 distributions @xmath570 and @xmath571 , will be given elsewhere .
here we just present a remarkable exact result for the characteristic function @xmath572 of the @xmath10-sites , joined uncutoff probability distribution of the spin boson occupations @xmath573 generalizing method of eq .
( [ theta2 ] ) for the @xmath10 sites , we find @xmath574 \frac{1}{\sqrt{\text{det}(g+z ) } } , \label{thetan}\ ] ] where @xmath339 is the correlation @xmath575-matrix @xmath567 , calculated in sect .
5.3 ; @xmath576 .
the characteristic functions in eqs .
( [ thetafinal ] ) and ( [ theta2final ] ) for the single - site and two - sites , joined distributions , calculated in sect .
5.2 , immediately follow from this general result if one sets all other @xmath577 , except @xmath578 or / and @xmath579 , to be zero , so that all irrelevant factors in eq .
( [ thetan ] ) become unity and only the determinants of the @xmath443- or @xmath455-matrices remain .
the joined occupations probabilities for the @xmath10 sites can be calculated simply by a differentiation , @xmath580 similar to the calculation in the end of sect .
5.2 . actually , due to the @xmath371-cutoff in eq .
( [ spintheta ] ) , only the first two occupation numbers , @xmath581 , are of interest .
note that the uncutoff distributions @xmath582 , @xmath215 in eqs .
( [ rhojacobi ] ) , ( [ rhotheta2 ] ) and @xmath583 contain , in fact , all effects of the constraints and spin interaction ( including the constraint interaction ) , except only the @xmath371-cutoff in eq .
( [ spintheta ] ) for the second step of the hilbert space reduction in eq .
( [ spinreduction ] ) , since they are calculated for the exact , constrained and @xmath371-cutoff , hamiltonian in eq .
( [ hising ] ) . the cutoff distributions @xmath584 , @xmath585 , and @xmath586 are simply their @xmath371-cutoff versions .
a crucial point is that the derived consistency equations ( [ g1j - consistency ] ) are the exact ones and , contrary to the mean - field theory equations , are valid both inside the entire critical region and outside it .
they are equations for the auto- and cross - correlations of neighboring spin bosons and can not be reduced to a simpler equation for the order parameter alone , as it was attempted in the original landau approach .
the latter is possible only approximately and only far outside the critical region into a well - ordered phase .
a solution of the derived exact equations ( [ g1j - consistency ] ) is more rich , than the mean - field one , and includes a fine structure and all details of the critical phenomena . at the wings of the critical region
these consistency equations describe a critical behavior with the critical exponents , different from the ones , predicted by the mean - field theory .
so , they yield the renormalization group results , but are based on the regular , quantum - field - theoretical analysis of green s functions , perfected to an exact rigorous analysis . moreover , the exact solution of the ising model , given by eq .
( [ g1j - consistency ] ) , describes both the mesoscopic and macroscopic systems and is valid not only for the asymptotics at the wings of the critical region , but also for the critical functions at a central part of the critical region .
the derived above two pairs of equations ( [ spinbetaeq])-([spingeq ] ) and ( [ spin1-contraction])-([spin2-contraction ] ) ( for the unconstrained , auxiliary coherent order parameter and green s functions of spin excitations and for the basis partial operator contractions , respectively ) together with a pair of equations ( [ spinbeta])-([sping ] ) for the true , constrained coherent order parameter and green s functions of spin excitations and eq . ( [ spinself - energy ] ) for the total irreducible self - energy constitute the exact system of equations for the magnetic phase transitions in the critical region .
it is crucial that they provide a fully regular , exact description of all critical phenomena and critical fluctuations , starting from the microscopic hamiltonian of spin interaction and the first principles of statistical physics .
they fully preserve a nonlinear , nonanalytical , critical structure of various statistical and thermodynamic quantities in the critical region , contrary to many other approximate or phenomenological models , which usually start from some unjustified , unreliable , ad hoc assumptions , hypotheses or simplifications in the hamiltonian or in the description , inconsistent with an actual critical behavior .
moreover , the derived exact equations have a canonical , universal , symmetric and highly branched structure .
all these features facilitate the appropriate controllable approximations , for example in the kernels of the integral equations , and finding the correct solutions for the critical region in a regular way , both for the mesoscopic and macroscopic systems .
for a discussion of the structure of these fundamental equations , we refer to sect .
3.5 , since it is very similar to the structure of the corresponding bec equations .
the main difference is a presence of many local constraint functions @xmath317 , instead of one global constraint @xmath49 in the bec problem , and , hence , many corresponding discrete variables @xmath587 in eqs .
( [ spin1-contraction])-([spin2-contraction ] ) for the basis partial operator contractions
. an important feature of the magnetic phase transitions in a system of spins is that the equations ( [ spin1-contraction])-([spin2-contraction ] ) for the basis partial operator contractions with a cutoff function @xmath588 can be solved exactly by a direct recurrence @xmath589 .
indeed , one starts from a zero boundary value @xmath590 at @xmath591 and ends at the required argument @xmath592 in just @xmath593 steps , depending on a spin value @xmath594 .
another important fact is that the derived equations get considerably simplified in the usually considered case of an interaction only between the neighboring spins in a homogeneous system .
these facts facilitate an exact solution of eq .
( [ spinself - energy ] ) for the self - energy in terms of a few exact correlation parameters for the coordination number @xmath425 of neighboring spins .
it yields a regular way for a solution of the critical phenomena problem .
namely that method is used for the exact solution of the long - standing 3d ising problem , outlined in sect .
the obtained exact system of @xmath595 consistency equations for the auto- and cross - correlations ( [ g1j - consistency ] ) is much more nontrivial , than the approximate mean - field self - consistency equation for the order parameter .
that results in a different critical behavior of the solution near the critical point and , in particular , in a deviation of the actual critical exponents for the 2d and 3d ising models from the ones , predicted by the mean - field theory .
specifically , the difference comes from a different behavior of the exact solution for the self - energies and eigen - energies near a singular point @xmath26 of the fourier sum or integral ( in a thermodynamic limit ) for the true correlation functions at a long range @xmath596 , @xmath597 on this basis , one can compare the long - range asymptotics of the latter integral @xmath598 , that is , the values of the corresponding critical exponent @xmath21 , within the present exact theory , wilson s @xmath0-expansion in the renormalization group theory , landau mean - field theory , and exactly solvable models .
the corresponding detailed analysis will be presented elsewhere .
the present regular theory can be compared against a few known solutions for the exactly solvable models @xcite .
the latter usually are unrealistic and demonstrate a degenerate behavior , barely related to a typical behavior of the actual physical systems .
an example is the 2d ising model for a square lattice of spins .
its solution was found by onsager @xcite in 1944 and by zamolodchikov @xcite in 1989 for the case of zero and non - zero external magnetic field , respectively . however , the most interesting point is that the present theory opens a regular way for a solution of the actual 3d problems ( see sect .
5 ) , for example , for the systems with the ising or heisenberg hamiltonians , which were never solved before .
an interesting insight of the rigorous theory is that a fundamental part in the process of the spin correlations and magnetic phase transition is playing by the spin - boson creation and annihilation operators in eq .
( [ spinhatbeta ] ) , rather than the spin operators themselves .
in fact , the mean values of the spin operators in eqs .
( [ spinvectorbeta])-([spin+-beta ] ) , that is a spontaneous magnetization of the spin lattice , should be calculated via the averages @xmath599 and @xmath600 .
the latter as well as the spin correlations and other statistical and thermodynamic quantities can be found from the coherent order parameter @xmath601 and green s functions of the spin bosons by means of the basis partial operator contractions similar to and based on the calculation of the true , constrained order parameter and green s functions in eqs .
( [ spinbeta])-([sping ] ) . obviously , an exact reduction of the many - body spin system to a constrained system with a purely bose statistics ( i.e. , with the pure bose commutation relations for the creation and annihilation operators of the system s excitations ) in sect .
4.1 means that the ferromagnetic and antiferromagnetic phase transitions in the lattice of spins are intimately related to those constraints .
usually an unconstrained bose system is not subject to a phase transition .
an example is the bec in a system of bose particles , which occurs only due to the particle - number constraint @xmath602 . in a system without that constraint , for example , in a system of photons in a black - box cavity with absorbing walls
, the bec phase transition does not exist .
thus , a fundamental symmetry , which control the dynamics and statistics of the many - body system of spins and which is generally wider than a symmetry , broken in a particular magnetic phase transition , is a local gauge symmetry of the spin bosons for each spin of the lattice .
( note that a similar , local rotational symmetry is the one broken in a nematic phase of the liquid crystals and in a superfluid @xmath603he . ) in virtue of the noether s theorem , the corresponding integrals of motion are the constraints in eq .
( [ spin ] ) , which determine , via the @xmath317-cutoff factors , the nontrivial structure of the spin excitations in eqs .
( [ spinalpha])-([spinhatbeta ] ) , the physical , constrained many - body hilbert space @xmath327 in eq .
( [ spinalpha=beta ] ) , and the hamiltonian in eqs .
( [ hxyhz])-([hising ] ) .
a spontaneous appearance of a macroscopic structure of magnetization in a ferromagnet or an antiferromagnet , that is a regular structure of the mean values of spins @xmath604 in the lattice , usually breaks only a subgroup of the fundamental symmetry group and does not correspond to any additional constraints .
the dynamics and statistics of the system is governed by those fundamental constraints , while one of the integrals of motion of the to - be - broken - symmetry subgroup could even coincide with an operator of an observable order parameter .
a mean value of the latter could appear and grow in a course of a phase transition from zero to a macroscopic value due to restructuring of a quantum - statistical state of the system .
this is what the magnetization does , for example , in the case of a homogeneous ferromagnetic phase transition .
hence , a magnetization , despite being a conserved quantity , may vary in mean value with changing temperature or other control parameters .
this is in contrast to the fundamental integrals of motion , which are the fixed c - number constraints on the total occupations ( [ spin ] ) of two spin bosons at each site of the lattice and always remain equal to twice the spin value , @xmath605 . in the case of bec in a gas ( sect .
3 ) there is only one such fundamental c - number constraint : the total occupation ( [ n ] ) of all single - particle energy levels in a trap is fixed to be equal to the total number of particles @xmath10 , loaded into a trap .
finally , the presented rigorous theory clearly proves that a used by many authors @xcite approximation of the nonanalytical functions @xmath317 and @xmath383 in the interaction hamiltonian ( [ hxyhz ] ) by the c - numbers or the polynomials is valid only asymptotically for the parameters far outside the critical region into a well - ordered phase .
it leads there to the landau mean - field theory or the dyson s spin waves theory and can not correctly describe the critical fluctuations and a transition from a disordered phase to an ordered phase through a critical point .
it is straightforward to extend the general method and structure of the microscopic theory of the critical phenomena , presented in @xcite and outlined in detail here , to other phase transitions in various fields of physics , including the physics of condensed matter and quantum fields .
the purpose of this paper is to bring an attention to the remarkable opportunity for a regular solution of the problem of phase transitions across a critical point , which is common to so many physical systems .
\(b ) the equation for the total irreducible self - energy @xmath128 , that determines a kernel in the first two equations ( a ) , @xmath375\delta\tilde{\bar{\beta}}_{j_2x_2 } \rangle = ( -1)^{j_1 } \int_0^{1/t } \sigma_{j_1x_1}^{jx}g_{jx}^{j_2x_2 } d\tau , \label{uself - energy}\ ] ] \(c ) the two equations for the basis partial one - operator ( eqs .
( [ s ] ) and ( [ spins ] ) ) and two - operator ( eqs .
( [ b ] ) and ( [ spinb ] ) ) contractions @xmath195= \bar{\beta}_j f(\ { m_{j ' } \})+ g_j^{j ' } \delta_{m_{j ' } } \tilde{s}_{j'}[f(\ { m_{j ' } \ } ) ] , \label{u1-contraction}\ ] ] @xmath196 = -g_{j_1}^{j_2}f(\ { m_{j ' } \ } ) - g_{j_1}^{j ' } g_{j'}^{j_2 } \delta_{m_{j'}}f(\ { m_{j ' } \})\ ] ] @xmath197 , \label{u2-contraction}\ ] ] \(d ) the following explicit formulas for the calculation of the true , constrained by the @xmath371-cutoff , order parameter and green s functions from the unconstrained ones in ( a ) and from the basis partial operator contractions in ( c ) : @xmath384 \rangle /p , \quad p=\langle \hat{\theta } \rangle , \label{ubeta'}\ ] ] @xmath606 + \tilde{b}^{j_2}_{j_1}[\tilde{\theta}_{\tau_1}\tilde{\theta}_{\tau_2 } ] \ } \rangle /p .
\label{ug'}\ ] ] the latter are similar to eqs .
( [ beta ] ) , ( [ g ] ) and ( [ spinbeta ] ) , ( [ sping ] ) .
all other statistical and thermodynamic quantities can be found from those unconstrained and true order parameters and green s functions in a similar way .
all these fundamental equations are written here in a short symbolic form that includes the particular cases of bec ( eqs .
( [ betaeq ] ) , ( [ geq ] ) and ( [ 1-contraction ] ) , ( [ 2-contraction ] ) ) and ferromagnetism ( eqs . ( [ spinbetaeq ] ) , ( [ spingeq ] ) and ( [ spin1-contraction ] ) , ( [ spin2-contraction ] ) ) .
formally , being considered for a given total irreducible self - energy @xmath128 separately from other equations , each of the eqs .
( [ ubetaeq ] ) , ( [ ugeq ] ) , ( [ u1-contraction ] ) , and ( [ u2-contraction ] ) is a linear equation .
however , together with eq .
( [ uself - energy ] ) for the self - energy @xmath128 they form , in fact , an essentially nonlinear system of equations since the self - energy @xmath128 ( see eqs .
( [ self - energy ] ) , ( [ spinself - energy ] ) ) is determined by their solutions and eqs .
( [ ugeq ] ) , ( [ u1-contraction ] ) , ( [ u2-contraction ] ) include the solutions of eqs .
( [ ubetaeq])-([ugeq ] ) as coefficients .
remarkably , these equations are valid in the entire critical region even for a mesoscopic system .
this result was achieved by formulating a theory at a more fundamental , operator level via an introduction of the recurrence eqs .
( [ u1-contraction])-([u2-contraction ] ) for partial operator contractions , which reproduce themselves under partial contraction operation .
various solutions of these exact equations and their application for various systems and models with the particular microscopic hamiltonians will be discussed elsewhere . in sect .
5 we outline them for a generic example of the exact solution to a long - standing 3d ising problem . the microscopic theory of phase transitions in the critical region and mesoscopic effects
are directly related to the numerous modern experiments on and numerical studies of the bec of a trapped gas ( including bec on a chip ) , where @xmath607 , ( see , for example , @xcite ) , superfluidity of @xmath608he in nanodroplets ( @xmath609 ) @xcite and porous glasses @xcite as well as various magnetic phase transitions @xcite , magnetic nanoparticles for information storage applications @xcite , metamagnetism @xcite , magnetic phase coexistence @xcite , interplay between magnetic phase transitions and bec in the spinor bose gases @xcite , nanomagnets @xcite , and so on . in particular , note that the rigorous analysis in sect .
4.1 also includes a matsubara - matsuda mapping of 1/2 spins onto a lattice of bosons @xcite as a particular approximate version of the holstein - primakoff representation .
it means that the presented microscopic theory of critical phenomena can be immediately applied also to the intensively studied , in the last two decades , bec in quantum xy - ordered magnets , since that bec is intimately related to the matsubara - matsuda mapping ( see a recent review ) . in conclusion , we present a rigorous microscopic theory of phase transitions across a critical point .
it includes the exact equations for the coherent order parameter and green s functions as well as the exact recurrence equations for the basis partial operator contractions .
the theory rigorously accounts for the local and global constraints of the many - body hilbert space , including the ones related to the spontaneously broken symmetry via the noether s theorem
. it also accounts for the nonpolynomial functions in the hamiltonian , including the ones originated from the constraints of the hilbert space ( the constraint nonlinear interaction ) . at last , using this theory , we outline the exact solution for the long - standing 3d ising model .
we especially emphasize the crucial role of the novel recurrence equations for the basis partial operator contractions , eqs .
( [ 1-contraction])-([2-contraction ] ) , ( [ spin1-contraction])-([spin2-contraction ] ) , ( [ u1-contraction])-([u2-contraction ] ) , which offer universal , powerful , and very convenient tools for the calculation of various quantities , describing the critical phenomena . as is shown , in the case of the 3d ising model in sect .
5 , this method is truly universal : one has just to select an appropriate , for a given quantity in question , function @xmath177 and solve these recurrence equations by the regular methods .
it is truly remarkable that , as is shown in the paper , the exact microscopic equations for such different phase transitions , as a bec one in an interacting gas and a ferromagnetic one in a lattice of spins , are completely similar .
frey and s. sun , _ `` magnetic nanoparticle for information storage applications '' , ch . 3 in `` inorganic nanoparticles : synthesis , applications , and perspectives '' , ed .
by c. altavilla , e. ciliberto _
( crc press , boca raton , london , 2011 ) . | the problem of finding a microscopic theory of phase transitions across a critical point is a central unsolved problem in theoretical physics . we find a general solution to that problem and present it here for the cases of bose - einstein condensation in an interacting gas and ferromagnetism in a lattice of spins , interacting via a heisenberg or ising hamiltonian . for bose - einstein condensation ,
we present the exact , valid for the entire critical region , equations for the green s functions and order parameter , that is a critical - region extension of the beliaev - popov and gross - pitaevskii equations . for the magnetic phase transition
, we find an exact theory in terms of constrained bosons in a lattice and obtain similar equations for the green s functions and order parameter . in particular , we outline an exact solution for the three - dimensional ising model
. published in physica scripta * 90 * , 108002 ( 2015 ) , doi:10.1088/0031 - 8949/90/10/108002 .
keywords : critical phenomena , mesoscopic system , phase transitions , spontaneous symmetry breaking , microscopic theory , bose - einstein condensation , ferromagnetism . |
kroto et al . ( 1985 ) proposed that the fullerene molecule c@xmath2 and related species play an important role in interstellar physics and chemistry .
this argument is based on the stability of the `` buckyball '' structure of fullerene compounds , as observed in the laboratory and confirmed by ab - initio calculations ( see also kroto 1987 , 1988 and references therein ) .
the formation mechanisms invoked to produce such spherical carbonaceous structures are usually based on reactions in circumstellar shells , with precursors such as acetylene and corannulene ( goeres and seldmayr 1992 ; kroto and jura 1992 ; bettens et al .
another way of obtaining fullerenes , together with polycyclic aromatic hydrocarbons ( pahs ) , is the photoerosion of larger hydrogenated amorphous carbon grains ( scott et al . 1997 ) .
fullerene or fullerane species have been searched for at various wavelengths in interstellar spectra .
they have also been proposed as explanations for many unidentified interstellar bands : the extinction bump at 2175 , the diffuse interstellar bands , the infrared emission features , and the extended red emission ( webster 1993a , b , c , 1996 ) .
the c@xmath2 uv absorption spectrum shows a band at 3860 ( heath et al .
snow and seab ( 1989 ) and somerville and bellis ( 1989 ) have searched for this feature without any clear evidence of its presence .
they derived a c@xmath2 upper limit of the order of @xmath30.01% of the cosmic carbon abundance in the diffuse medium , and approximately @xmath30.7% of the carbon abundance in mira stars . more recently ,
clayton et al . (
1995 ) have searched for the infrared ( ir ) emission feature of c@xmath2 at 8.6 @xmath0 m in the spectra of r coronae borealis stars .
no feature was seen in these carbon - rich , hydrogen - deficient stars to a level of 2% of the continuum . according to these authors
, only the evolved carbon star irc+10216 possibly shows a 8.6 @xmath0 m band that may be assigned to c@xmath2 . what is the dominant ionization state of c@xmath2 in the interstellar environment ?
we consider recent ionization models for pahs ( bakes and tielens 1995 ; salama et al .
1996 ; dartois and dhendecourt 1997 ) .
the comparison between pahs and c@xmath2 is relevant , as the ionization potentials for c@xmath2 and pahs of the same size are similar ( see table 2 in leach 1995 ) .
the models conclude that a 60 atom pah would be mostly neutral ( pah ) and anionic ( pah@xmath4 ) in the diffuse medium , and cationic ( pah@xmath5 ) in more irradiated clouds such as reflection nebulae near a hot star .
only the model of bakes and tielens takes into account the second ionization stage ( pah@xmath6 ) . in the standard pdr model
they adopt , 50% of the 60 atom species would be singly ionized , 35% would be neutral and 15% doubly ionized . in the case of ngc 7023
, the radiation field is 10 times fainter than in this model , so that the fraction of c@xmath7 is even lower and we shall ignore it hereafter . laboratory work by petrie et al .
( 1993 ) has shown that c@xmath1 is the most stable species among 60 atom fullerenes , because of its `` exceptional unreactivity '' with other interstellar material .
the cation c@xmath1 has been proposed as a diffuse interstellar band ( dib ) carrier ( lger et al .
it possesses electronic modes at wavelengths shorter than 1 @xmath0 m , identified by laboratory spectra in rare - gas matrices ( fulara et al .
1993 ) , which were expected to be observed in absorption in the interstellar medium .
foing and ehrenfreund ( 1994 , 1997 ) have observed absorption features at 9577 and 9632 along a few lines of sight , which could correspond to these modes .
this assignment of the two near - ir bands to c@xmath1 , however , is debated ( maier 1994 , jenniskens et al .
1997 ) and we will discuss this point carefully in section 2 .
the c@xmath1 species would re - emit any energy absorbed from a neighboring star through its four ir modes .
two of these modes have been measured in the laboratory to fall at 7.11 and 7.51 @xmath0 m ( fulara et al . 1993 ) .
the goal of this paper is to search for the vibrational signature of c@xmath1 in the high signal - to - noise spectrum of the reflection nebula ngc 7023 .
we will also search for c@xmath2 bands , measured in the laboratory at 7.0 and 8.45 @xmath0 m ( krtschmer et al . 1990 ) .
the laboratory spectrum of c@xmath1 shows two electronic bands at 9583 and 9645 ( 10435 and 10368 @xmath8 ) in a neon matrix ( fulara et al .
the corresponding absorption bands have been searched for in the interstellar medium ( ism ) , taking into account the wavelength shift induced by the matrix polarisability .
the two new interstellar bands observed at 9577 and 9632 ( 10442 and 10382 @xmath8 ) by foing and ehrenfreund ( 1994 , 1997 ) have been tentatively assigned to the cation fullerene c@xmath1 . to deduce the abundance from the observations
requires an estimation of the oscillator strength .
the most reliable estimates ( @xmath9 = 0.003 0.006 from fulara et al .
1993 ) lead to a total abundance of 0.6 to 1.2% of cosmic carbon in this species alone ( moutou et al .
1996a ) , if the c / h ratio is taken as the solar value ( 3.7 @xmath10 10@xmath11 ) . with a more recent estimate of the interstellar c / h value ( 2.6 @xmath10 10@xmath11 , cardelli et al .
1996 ) , c@xmath1 would contain 0.85 1.7% of carbon .
this is significant in comparison with the estimated abundance of pahs in the interstellar medium , which is @xmath1210 15% of cosmic carbon ( joblin et al .
1992 , sellgren et al .
1995 , jones et al .
1996 , dwek et al .
1997 ) .
more recent observations of the 9577 and 9632 dibs ( jenniskens et al .
1997 ) lead to a lower value of the measured equivalent width , and thus to a lower abundance range for c@xmath1 : 0.25 to 0.52% of interstellar carbon ( with c / h from cardelli et al .
the reason for this disagreement is a different evaluation of the telluric contribution in the dib profiles .
maier ( 1994 ) and jenniskens et al .
( 1997 ) are both very cautious concerning the c@xmath1 identification , on the basis of : i ) the absence of weaker bands in the ism near 9400 , which are expected from the laboratory spectrum ; ii ) the difference in the intensity ratio of the two bands in the laboratory and in interstellar spectra ; and iii ) the larger shift in wavelength from the neon matrix to the ism compared to the wavelength shift between rare - gas matrices and the gas phase measured in similar molecules . as an attempt to clarify this situation
, we propose an alternate determination of the c@xmath1 abundance through its vibrational emission bands .
ionization models ( bakes and tielens 1995 ) predict the neutral c@xmath2 abundance to be of the order of 0.7 times the c@xmath1 abundance .
the ir laboratory spectrum of c@xmath1 has been measured in a neon matrix ( fulara et al .
1993 ) and exhibits two bands at 7.11 and 7.51 @xmath0 m ( 1406 and 1331 @xmath8 ) .
we reproduce the laboratory spectrum in figure 2 , courtesy of the authors .
two other bands are expected for symmetry reasons , but they have not been measured so far because their wavelengths are longer than the 16.7 @xmath0 m ( 600 @xmath8 ) cutoff of the fulara et al . experiment .
the c@xmath1 features have widths of 0.05 @xmath0 m ( 10 @xmath8 ) in the neon matrix at low temperature .
the ratio of the integrated cross - section of the 7.11 @xmath0 m band to the integrated cross - section of the 7.51 @xmath0 m band is measured to be 2.2 ( fulara et al . 1993 ) .
the modes of neutral c@xmath2 are already known to lie at 7.0 and 8.45 @xmath0 m ( 1435 and 1190 @xmath8 ; krtschmer et al .
temperature - dependence studies have shown that at 1000 k a shift of @xmath1220 @xmath8 could be expected ( nemes et al .
1994 ) , compared to the cold neon matrix . in a similar way , joblin et al .
( 1995 ) have shown that the widths of pah features measured in emission at high temperature in the laboratory are broadened by 0 to 0.1 @xmath0 m
( 0 to 20 @xmath8 ) compared to absorption bands in a 4 k neon matrix .
they also emphasize that the observed wavelengths of pah features shift to longer wavelength with increasing emission temperature .
the amount of the wavelength shift varies between different vibrational modes of a pah , but the redshift is in the range of 0 to 0.1 @xmath0 m ( 0 to 20 @xmath8 ) compared to pahs at low temperature .
no experimental studies have been published for ionic pah species .
it can nevertheless be expected that c@xmath1 features are broadened and shifted by 0 to 20 @xmath8 as well .
the two c@xmath1 features we wish to detect are therefore expected to lie at 7.1 7.2 @xmath0 m and at 7.5 7.6 @xmath0 m .
we have considered here two cases which are illustrated in figure 3 . in case a , we assume that other carbonaceous species , such as pahs or very small grains , emit the 6.2 , 7.7 , and 8.6 @xmath0 m emission features , and that the combined emission of these features can be treated as a local continuum
. we will not address here the nature of these emitters . for case a
, we have searched for c@xmath1 features at 7.1 7.2 @xmath0 m and at 7.5 7.6 @xmath0 m as weak features above the local continuum defined by the blue side of the broad 7.7 @xmath0 m feature profile . in case b , we assume that the broad 6.2 , 7.7 , and 8.6 @xmath0 m features are each made of many overlapping bands from individual aromatic molecules , and that two of the bands contributing to this emission complex could be due to c@xmath1 . the deduced abundance for c@xmath1 would be dramatically different between these two cases , as is shown in figure 3 . for both case
a and case b , we adopt gaussian profiles with a fwhm of 0.05 @xmath0 m for the two c@xmath1 bands , and we place the constraint on any possible c@xmath1 features that the ratio of the integrated cross - section of the 7.11 @xmath0 m band to the integrated cross - section of the 7.51 @xmath0 m band equal the laboratory value of 2.2 ( fulara et al .
this constraint limits the maximum contribution of c@xmath1 , by requiring that in case b the stronger of the two c@xmath1 features not exceed the total observed flux at 7.11 @xmath0 m in ngc 7023 .
the exciting star of the reflection nebula ngc 7023 is the be star hd 200775 , which has an effective temperature of 17000 k ( strom et al . 1972 ) . for detailed reviews of previous studies of this object , see lemaire et al .
( 1996 ) and martini et al .
our sws observations are higher spectral resolution than previously published mid - ir spectra of ngc 7023 from the kuiper airborne observatory ( sellgren et al .
1985 ) and from iso with isophot - s and isocam ( laureijs et al . 1996 ; cesarsky et al . 1996 ) .
the data presented here were taken by the sws spectrometer on board the iso satellite ( kessler et al . , 1996 ) during revolution 455 ( february 13th , 1997 ) .
we used the partial grating scan mode ( aot6 , band 2c ) of the sws instrument ( de graauw et al .
the position angle of the major axis of the sws rectangular aperture ( 14@xmath13 @xmath10 20@xmath13 for the wavelength range observed here ) was 258.5@xmath14 .
the aperture is thus located at the southern edge of the northwestern filament observed by isocam ( as seen in the 6.2 to 11.3 @xmath0 m feature ratio , cesarsky et al . , 1996 ) and is parallel in orientation to this filament . in the sws grating spectrometer , light is dispersed along the major axis .
data gathered by the sws spectrometer have a high redundancy : 12 detectors and two grating scan directions ( up and down hereafter ) , for a total of 24 individual spectral scans .
there are two grating steps per spectral resolution element ( @xmath15 m ) ; each spectral resolution element is thus observed independently 48 times . to reduce the data
, we used the current sws - ia version running at the institut dastrophysique spatiale ( last update : october 1997 ) .
the data reduction steps are summarized in de graauw et al .
the flux calibration was done with the file cal25 - 2c-020 . in the following
, we detail our modifications to the standard reduction scheme in order to improve the signal - to - noise ( s / n ) ratio .
first , we removed glitches due to cosmic rays from the spectra . for the band 2c detectors ,
glitches rarely affect more than a single read - out ( 2 seconds ) . to detect glitches ,
we subtract a continuum ( a sliding boxcar on 15 points , with 2-@xmath16 outliers rejected ) from each of the 24 individual scans , to create individual high - frequency ( hf ) scans .
the individual hf scans , which now contain unresolved lines , glitches and noise , are combined by wavelength .
the combined hf spectrum has 48 points per resolution element ; we smooth it with a 15-point sliding boxcar to improve our s / n on the combined hf signal while retaining enough spectral sampling to detect true hf features in the spectrum such as unresolved emission lines .
we subtract the combined hf signal from each hf individual scan , leaving individual residual scans which in principle only contain noise and glitches .
we finally remove glitches from each individual residual scan by 3-@xmath16 clipping , with @xmath16 determined locally from a 10-point sliding boxcar . between 10 and 20% of the data points were identified as glitches .
second , we removed correlated noise , due to weaker glitches ( not necessarily due to cosmic rays like , e.g. , electrical noise ) which occur in all 12 detectors at the same time .
correlated noise is not always detected by the above glitch removal technique , and can result in a false spectral feature which appears broader than the spectral resolution ( see sellgren et al .
1999 ) . to eliminate correlated noise , we subtracted a continuum from each scan to create individual hf scans , as described above , but then combined the hf scans in the time domain ( at each reset point ) rather than in the wavelength domain .
we next subtracted from each individual hf scan the mean hf signal ( after sigma - rejection ) of all 12 detectors taken at each reset point .
the individual continuum is subsequently restored into each corrected hf scan .
unresolved lines remain unaffected by this method as our sampling is dense enough . by applying this signal decorrelation
, we reduced the rms noise by 5 to 15% .
the two operations described above , to correct for glitches and correlated noise , were performed just after flux calibration . as they were performed on the hf part of the signal
they do not depend on flatfield quality .
the third step in the data reduction is the flatfielding .
we used the standard sws flatfielding routine to scale all individual scans to a reference .
the reference is defined by a linear fit to all 12 down scans .
the scaling factor for each individual scan is defined as the reference divided by a linear fit to the individual scan . if an individual scan has some fit points close to zero , the scaling factor becomes large and the amplification of noise in the scan adversely affects the flatfield quality . to avoid this
, we added a constant value ( 100 jy ) to all individual scans , performed the flatfield correction ( multiplication by the scaling factors ) , then subtracted this constant value to produce the final flux level .
the final average spectrum was produced with a sliding boxcar on a given binsize ( 0.0042 @xmath0 m or mean resolving power of 1900 , valentijn et al . , 1996 ) .
in order to be sure that the residual error includes any true differences between the up and down scans , we performed a second reduction in which we separated the up and down scans just before the flatfield operation .
the two scan directions were then flatfielded completely independently .
this was used only to calculate the residual error ( up @xmath17 down)/2 .
we will now derive an upper limit on the c@xmath1 and c@xmath2 abundances , for both case a and case b. for this purpose , we compare the observed spectrum of ngc 7023 to a calculated emission spectrum of c@xmath1 .
this is more rigorous than extracting a relative value of c@xmath1 emission compared to the pah emission , especially in the 7 8 @xmath0 m domain , where the spectroscopic match of pahs to the interstellar data is not yet proved .
we will first show the detailed calculation for c@xmath1 , and then apply it to c@xmath2 .
the c@xmath1 absorption spectrum we adopt for modeling purposes reproduces the bandwidths and relative intensities measured in laboratory experiments ( fulara et al .
we use the wavelengths and relative intensities of the long - wavelength modes in c@xmath2 ( at 17.3 and 18.9 @xmath0 m ) to approximate these modes in c@xmath1 , since the long - wavelength modes of c@xmath1 are predicted but not yet measured . we assume the wavelengths of the c@xmath1 modes do not depend on temperature , because we can not predict this dependence ( although any wavelength shift is probably less than 0.1 @xmath0 m ) .
it is important to note that the power emitted in the c@xmath1 features does not depend on the absolute value of the ir cross - section .
the reason for this is energy conservation ; the total absorbed power is equal to the total emitted power , independent of the cooling time between uv photons .
according to the thermal model ( sellgren 1984 , lger et al .
1989 ) , the excitation of a species after a single uv or visible photon absorption raises its vibrational temperature to a high value , determined by its size via the heat capacity .
the emission results from the cooling by vibrational decay . to calculate the ir emission spectrum of c@xmath1
, we thus have to model its absorption of stellar light and its emission in the 7 @xmath0 m bands . the total power absorbed by c@xmath1 molecules near a star is given by : @xmath18 where @xmath19 is the illuminating stellar flux as a function of frequency @xmath20
, @xmath21 is the uv absorption cross - section per carbon atom for a c@xmath1 molecule , and @xmath22 is the number of carbon atoms in the observed source which are contained within c@xmath1 molecules .
the uv absorption cross - section of the c@xmath1 cation is not directly known from laboratory experiments .
we therefore use the value measured from experiments on neutral pahs , normalized to one carbon atom , as it seems to be similar for graphite and pahs and thus characteristic of sp@xmath23 hybridized carbon ( joblin et al .
1992 , verstraete & lger 1992 ) .
we note that the absorption cross - section for an ionized species is expected to be larger than for a neutral one , especially in the visible ( leach 1987 ) .
our assumption thus probably underestimates the actual absorption , which is suitable for determining an upper limit of the species abundance .
the emitted spectral energy distribution , @xmath24 , depends on the infrared absorption cross - section as a function of wavelength , and the details of how the molecule cools after absorption of a visible or uv photon , as described by lger et al .
we define @xmath25 , the fraction of the total c@xmath26 emitted energy which is emitted in the c@xmath26 bands at 7.11 and 7.51 @xmath0 m , as @xmath27 where the frequency limits @xmath28 and @xmath29 indicate an integration just over the 7.11 and 7.51 @xmath0 m bands .
the total c@xmath26 emitted energy will include emission in the c@xmath26 bands at 7.11 and 7.51 @xmath0 m , emission in the c@xmath26 long - wavelength modes ( approximated by the c@xmath2 modes at 17.3 and 18.9 @xmath0 m ) , and possibly also some unknown amount of fluorescence .
such fluorescence is observed in some neutral aromatic species ( negri & zgierski 1994 ) .
the c@xmath1 fluorescence quantum yield is not known , and so we assume it is negligible . including
a finite amount of c@xmath1 fluorescence would have the effect of increasing our upper limit on the c@xmath1 abundance .
our calculations , ignoring fluorescence , find that the fraction of the total emitted energy from c@xmath1 is 35% at 7.11 @xmath0 m , 15% at 7.51 @xmath0 m , 10% at 17.3 @xmath0 m , and 40% at 18.9 @xmath0 m .
this gives @xmath25 = 50% as the fraction of total c@xmath1 emitted energy resulting from emission at 7.11 and 7.51 @xmath0 m .
energy conservation requires that @xmath30 = @xmath31 . to convert the emitted power spectrum from c@xmath1 to an observed total intensity spectrum , @xmath32
, we compare the power absorbed and re - emitted in the c@xmath1 bands to the power absorbed and re - emitted by dust grains , @xmath33 and @xmath34 .
we derive @xmath34 from the integration of the sws01 and lws01 spectra of ngc 7023 ( sellgren et al .
1999 ) , over the wavelength range 3 200 @xmath0 m .
our value of i@xmath35 given in table 1 , 1.0 @xmath10 10@xmath36 w m@xmath37 sr@xmath38 , is compatible with previous measurements ( whitcomb et al .
1981 ) .
the power thermally emitted by the entire grain population is related to the total stellar flux absorbed by dust by : @xmath39 where @xmath40 is the total number of hydrogen atoms , @xmath41 is the grain extinction cross - section normalized to one hydrogen atom , and @xmath42 is the grain albedo .
we adopt the extinction curve of the diffuse interstellar medium and the associated albedo as given in dsert et al .
we integrated over ten points in the uv - visible range .
the fraction of carbon locked up in c@xmath1 , @xmath43 , is then given by : @xmath44 the first term in this equation is @xmath45 , the fraction of total c@xmath1 energy emitted at 7.11 and 7.51 @xmath0 m .
the second term is the ratio of stellar flux absorbed per hydrogen atom by dust to the stellar flux absorbed per carbon atom by c@xmath1 .
this term is effectively the ratio of the dust absorption cross - section to the c@xmath1 absorption cross - section , weighted by the stellar flux .
the stellar flux is approximated by a black body emission spectrum at the effective temperature of hd 200775 , i.e. 17,000 k. the third term is the interstellar carbon abundance , @xmath46 , as measured by cardelli et al .
( 1996 ) in the local interstellar medium .
the last term is our observed upper limit on the 7.11 and 7.51 @xmath0 m c@xmath1 emission , @xmath47 , divided by the observed total infrared emission from dust , @xmath34 .
our adopted values for all these terms used in equation 4 are summarized in table 1 .
.summary of calculated quantities and observables which lead to an upper limit on the c@xmath1 abundance in ngc 7023 . [ cols= " < , < " , ] let us consider case a , as presented in the top panel of figure 3 , where the two c@xmath1 bands are small bumps on top of a local continuum .
we adopt a second - order local continuum and we fit each small bump with a gaussian of 0.05 @xmath0 m fwhm .
the residual emission at 7.11 and 7.51 @xmath0 m , after continuum subtraction , is only significant at the 2.5-@xmath16 level and thus we consider this to be an upper limit on the maximum integrated energy of any c@xmath1 features .
we estimate this upper limit to be @xmath32.0 @xmath10 10@xmath48 w m@xmath37 sr@xmath38 for the sum of both bands .
the deduced upper limit for case a is @xmath30.027% of carbon atoms contained in c@xmath1 towards ngc 7023 ( equation 4 ) .
we now consider the other extreme , case b , where we place an upper limit on c@xmath1 as illustrated in figure 3b .
we again adopt gaussians of width similar to those measured in the laboratory ( fwhm @xmath49 0.05 @xmath0 m ) , for both the 7.11 and 7.51 @xmath0 m bands , and require that the ratio of the integrated cross - section of the 7.11 @xmath0 m band to the integrated cross - section of the 7.51 @xmath0 m band equal the laboratory value of 2.2 ( fulara et al .
1993 ) . these requirements for our case b
upper limit imply that all the emission at 7.11 @xmath0 m , but only part of the emission at 7.51 @xmath0 m , is due to c@xmath1 in the observed spectrum of ngc 7023 . for case b , the upper limit on the total energy emitted by c@xmath1 at 7.11 and 7.51 @xmath0 m is @xmath31.9 @xmath10 10@xmath50 w m@xmath37 sr@xmath38 .
this leads to an upper limit on the c@xmath1 abundance of @xmath30.26% of the interstellar carbon in ngc 7023
. our case b method gives an upper limit on the c@xmath1 abundance ( @xmath30.26% of interstellar carbon ) which is just compatible with the c@xmath1 abundance ( 0.25 0.52% of interstellar carbon ) inferred from the dibs observed at 9577 and 9632 by jenniskens et al .
our case b results , however , are significantly lower than the c@xmath1 abundance ( 0.85 1.7% of interstellar carbon ) inferred from the dib observations of foing and ehrenfreund ( 1994 , 1997 ) .
our upper limit on the c@xmath1 abundance from case a ( @xmath30.027% of interstellar carbon ) is a factor of ten below the c@xmath1 abundances inferred from any of the dib observations .
we note that the identification of the 9577 and 9632 dibs with c@xmath1 , however , is still debated ( foing and ehrenfreund 1994 , 1997 ; maier 1994 ; jenniskens et al .
we also note that if the c@xmath1 fluorescence quantum yield , @xmath51 , is not negligible , it would raise both our case a and case b upper limits by a factor of ( 1 @xmath17 @xmath51)@xmath38 . in case b , we assume that c@xmath1 is one of many narrow ( fwhm @xmath12 0.05 @xmath0 m ) aromatic bands blended together to emit features at 6 9 @xmath0 m , as has been proposed in the pah model ( lger & puget 1984 , allamandola et al .
if this view of the origin of the interstellar emission features at 6 9 @xmath0 m is correct , then this has the interesting implication that @xmath52 different aromatic species would be required to account for the entire 6 9 @xmath0 m emission complex .
it is not clear , however , whether such a mix of aromatic molecules could also account for the underlying smooth continuum at 1
14 @xmath0 m , particularly in wavelength regimes where pahs do not have vibrational transitions .
our case b upper limit on the c@xmath1 abundance is set simply by the total observed flux at 7.11 @xmath0 m in ngc 7023 . if there are additional aromatic molecules which contribute to the observed flux at 7.11 @xmath0 m , as seems likely , then our upper limit to the c@xmath1 abundance of @xmath30.26% of interstellar carbon is an overestimate .
we also searched for the neutral c@xmath2 features near 7.0 7.1 and 8.45 8.6 @xmath0 m ( krtschmer et al .
1990 , nemes et al .
we found no resolved emission features at the expected wavelengths .
we estimate an upper limit on the neutral c@xmath2 abundance using the case b method , by assuming that all of the the observed flux in ngc 7023 at 7.0 @xmath0 m is due to c@xmath2 .
we require that the ratio of the integrated cross - section of the 8.45 @xmath0 m band to the integrated cross - section of the 7.0 @xmath0 m band matches the experimental ratio of 0.75 , and we assumed that the features were gaussians with a width of 0.05 @xmath0 m .
we find an upper limit on @xmath53 ( case b ) of @xmath32.0 @xmath10 10@xmath50 w m@xmath54 .
we follow the same procedure for c@xmath2 as for c@xmath1 , as our calculation did not make any corrections for the ionization state of the fullerene .
this implies a case b upper limit for the abundance of the neutral c@xmath2 molecule of of @xmath30.27% of interstellar carbon .
ionization models predict that the neutral c@xmath2 abundance should be @xmath120.7 times the c@xmath1 abundance in the highly illuminated environment of a b star ( bakes and tielens 1995 ) .
our case b upper limits on the abundances of c@xmath2 and c@xmath1 , respectively @xmath30.27% and @xmath30.26% of interstellar carbon , do not contradict the predictions of ionization models , as we do not detect either molecule in ngc 7023 .
we have searched for the vibrational emission bands of c@xmath1 , at 7.11 and 7.51 @xmath0 m , and of c@xmath2 , at 7.0 and 8.45 @xmath0 m , using a new iso spectrum of the reflection nebula ngc 7023 .
we do not detect emission from either fullerene molecule , and we place upper limits on their abundances .
our upper limit on the c@xmath1 abundance , from the lack of observed vibrational emission , is independent of other estimates based on assigning diffuse interstellar bands at 9577 and 9632 to electronic absorption bands of c@xmath1 ( jenniskens et al .
1997 , foing & ehrenfreund 1994 , 1997 ) .
if we assume that all of the flux observed toward ngc 7023 at 7.11 @xmath0 m is due to c@xmath1 , then we derive an upper limit on the c@xmath1 abundance of @xmath30.26% of interstellar carbon toward ngc 7023 .
this upper limit is just compatible with the jenniskens et al .
( 1997 ) observations , which imply 0.25 to 0.52% of interstellar carbon is in c@xmath1 , and is significantly below the c@xmath1 abundance inferred from foing & ehrenfreund ( 1994 , 1997 ) . if we instead assume that c@xmath1 would produce only a small bump on a local continuum due to the blue wing of the 7.7 @xmath0 m interstellar emission feature , then our upper limit on the c@xmath1 abundance , @xmath30.027% of interstellar carbon , is a factor of ten less than the estimates from the 9577 and 9632 dibs .
the apparent lack of c@xmath1 toward ngc 7023 is unlikely to be due to the ionization state of the fullerene .
our upper limit on c@xmath2 , @xmath30.27% of interstellar carbon , is comparable to our upper limit on c@xmath1 .
ionization models of aromatic hydrocarbons predict that the fraction of c@xmath7 is negligible in ngc 7023 ( bakes & tielens 1995 ) .
these models also predict that the c@xmath1 fraction should be higher in ngc 7023 than in the diffuse medium where foing & ehrenfreund ( 1994 , 1997 ) and jenniskens et al .
( 1997 ) made their observations .
there are several possible explanations for why c@xmath1 is not detected toward ngc 7023 in the amount inferred from the near - ir dib observations .
it is possible that we have overestimated the fraction of energy absorbed by c@xmath1 which is re - radiated near 7 @xmath0 m , for instance if the c@xmath1 far - infrared modes are stronger than expected from data on c@xmath2 , or if fluorescence is non - negligible . also , while the 9577 and 9632 dibs have not yet been observed toward the central star in ngc 7023 , other dibs
are observed to be anomalously weak toward this star ( snow et al .
1995 ; oudmaijer et al .
jenniskens et al .
( 1997 ) observe that the 9577 and 9632 dibs follow the behavior of other dibs ( jenniskens et al .
1994 ) in being significantly fainter in dense molecular clouds with strong uv irradiation .
a third explanation is that the 9577 and 9632 dibs are not due to c@xmath1 , as argued by maier ( 1994 ) and jenniskens et al .
( 1997 ) .
new laboratory data will be critical to answering some of these questions .
it is essential to obtain the near - ir gas - phase spectrum ( 0.94 to 1.0 @xmath0 m ) of c@xmath1 , in order to confirm or deny the assignment of the 9577/9632 dibs to c@xmath1 .
it is important to measure the broadening and wavelength shift induced by temperature effects on the vibrational bands of c@xmath1 , which we can only estimate from laboratory measurements on similar molecules
. it would also be useful to observe the full - range ir absorption spectrum of c@xmath1 to measure the relative cross - sections of all four vibrational bands .
we conclude that c@xmath2 related species do not appear to be a dominant form of carbonaceous species in the interstellar medium .
confirmation of this result , as well as further searches for fullerene or fullerane species in interstellar spectra , should be pursued at various wavelengths .
additional searches for the fullerene bands near 7 @xmath0 m in circumstellar environments , where c@xmath2 and related molecules could be formed , could be particularly fruitful .
+ * acknowledgements * : the infrared space observatory is a project of the european space agency .
we particularly thank the sws operational team at vilspa for their efforts to provide the calibration and data reduction systems .
we gratefully acknowledge nato collaborative research grant 951347 .
allamandola , l.j . ,
tielens , a.g.g.m . , & barker , j.r .
, 1985 , apjl 290 , l25 bakes e.l.o .
and tielens a.g.g.m .
, 1995 , `` the diffuse interstellar bands '' , eds .
aggm tielens and tp snow , p.315 bettens r.p.a . &
herbst e. 1997 , apj 478 , 585 cardelli j.a .
, meyer d.m . , jura m. & savage b.d . , 1996 ,
apj 467 , 334 cesarsky .
d. , lequeux j. , abergel a. et al .
, 1996 , a&a 315 , l305 clayton g.c . ,
kelly d.m .
, lacy j.h . , little - marenin i.r .
, feldman p.a . &
bernath p.f .
1995 , aj 109 , 2096 dartois e. and dhendecourt l. , 1997 , a&a 323 , 534 dsert f.x .
, boulanger f. & puget j.l .
, 1990 , a&a 237 , 215 de graauw t. , haser l.n .
, beintema d.a .
et al . , 1996 , a&a 315 , l49 dwek e. , arendt r.g . , fixen d.j .
et al . , 1997 , apj 475 , 565 foing b. and ehrenfreund p. , 1994
, nature 369 , 296 foing b. and ehrenfreund p. , 1997
, a&a 317 , l59 fulara j. , jakobi m & maier j.p . , 1993 , chem .
letters 211 , 227 goeres , a. and seldmayr , e. 1992 , a&a 265 , 216 heath j.r . , curl r.f . & smalley r.e .
1987 , j. chem .
87 , 4236 jenniskens p. , ehrenfreund p. and foing b. , 1994 , a&a 281 , 517 jenniskens p. , mulas g. , porceddu i. & benvenuti p. 1997 , a&a 327 , 337 joblin c. , lger a. , martin p. , 1992 , apj 393 , l79 joblin c. , boissel p. , lger a. , dhendecourt l. , dfourneau d. , 1995 , a&a 299 , 835 jones a.p .
, tielens a.g.g.m . &
hollenbach d.j . , 1996 , apj 469 , 740 kessler m. , et al .
1996 , a&a 315 , l27 krtschmer w. , lamb l.d . & fostiropoulos k. , 1990 , nature 347 , 354 kroto h.w . , heath j.r . , obrien s.c .
, curl r.f . &
smalley r.e . , 1985 ,
nature 318 , 162 kroto h.w .
, 1987 , in `` pahs and astrophysics '' , eds .
a. lger , l. dhendecourt and n. boccara , p.27 kroto h.w . , 1988 ,
science 242 , 1139 kroto h.w & jura m. , 1992 , a&a 263 , 275 laureijs r.j . , acosta - pulido j. , abraham p. et al .
1996 , a&a 315 , l313 leach s. 1987 , in `` pahs and astrophysics '' , eds .
a. lger , l. dhendecourt and n. boccara , p.99 leach s. , 1995 , in `` the diffuse interstellar bands '' , eds . a.g.g.m .
tielens and t.p .
snow , p. 281 lger , a. , puget , j.l .
1984 , a&a 137 , l5 lger a. , dhendecourt l. , verstraete l. & schmidt w. 1988 , a&a 203 , 145 lger a. , dhendecourt l. & dfourneau d. 1989 , a&a 216 , 148 lemaire j.l . , field d. , gerin m. , leach s. , pineau des forts rostas , f. & rouan d. , 1996 , a&a 308 , 895 maier j.p .
1994 , nature 370 , 423 martini p. , sellgren k. & hora j.l . , 1997 , apj 484 , 296 moutou c. , lger a. , dhendecourt l. & maier j. , 1996a , a&a 311 , 968 negri , f. & zgierski , m.z . , 1994 , j. chem phys 100 , 1387 nemes l. , ram s.r . , bernath p.f
, et al . , 1994 , chem .
physics letters 218 , 295 oudmaijer r.d .
, busfield , g. , & drew , j.e .
1997 , mnras 291 , 797 petrie s. , javahery g. & bohme d.k . , 1993 ,
a&a 271 , 662 salama f. , bakes e.l.o , allamandola l.j .
& tielens a.g.g.m . , 1996 ,
apj 458 , 621 scott a. , duley w.w . & pinho g.p . , 1997 , apj 489 , l193 sellgren k. , 1984 , apj 277 , 623 sellgren k. , allamandola l.j .
, bregman j.d . , werner m.w . &
wooden d.h . , 1985 , apj 299 , 416 sellgren k. , brooke t. y. , smith r. g. , & geballe t. r. , 1995 , apjl 449 , l69 sellgren k. et al .
1999 , in preparation snow t.p .
, bakes e.l.o .
, buss r.h . , &
1995 , a&a 296 , l37 snow t.p . &
1989 , a&a 213 , 291 somerville w.b .
& bellis j.g .
1989 , mnras 240 , 41 strom s.e .
, strom k.m . , yost j. , carrasco l. , grasdalen g. , 1972 , apj 173 , 353 valentijn e.a . , feuchtgruber h. , kester d.j.m et al
. , 1996 , a&a 315 , l60 verstraete l. and lger a. , 1992 , a&a 266 , 513 webster a. , 1993a , mnras 263 , l55 webster a. , 1993b , mnras 264 , 121 webster a. , 1993c , mnras 264 , l1 webster a. , 1996 , mnras 282 , 1372 whitcomb s.e .
, gatley i. , hildebrand r.h . , keene j. , sellgren k. & werner m.w . , 1981 , apj 246 , 416 | we present here the 7.0 8.7 @xmath0 m spectrum of the bright reflection nebula ngc 7023 .
our observations are made with the short wavelength spectrometer ( sws ) on the european satellite infrared space observatory ( iso ) .
the vibrational bands of the ionized fullerene c@xmath1 are expected at 7.11 and 7.51 @xmath0 m , while those of the neutral fullerene c@xmath2 are expected at 7.0 and 8.45 @xmath0 m .
we estimate an upper limit in ngc 7023 for the c@xmath1 abundance of @xmath30.26% of the interstellar abundance of carbon , while c@xmath2 contains @xmath30.27% of interstellar carbon . |
the need to sample random vectors of correlated binary variables arises in various statistical application ; examples are estimation of the posterior mean in bayesian variable selection @xcite , small - sample properties of estimators in longitudinal studies ( * ? ? ?
* for a recent review ) , stochastic binary optimization in combinatorics @xcite , simulation of ferromagnetic materials @xcite , performance of neural networks @xcite and market segmentation analysis @xcite among others .
let @xmath0 denote the binary space . in some cases , such as small - sample analysis in longitudinal studies ,
we need a parametric family @xmath1 explicitly for sampling data on @xmath2 with specified mean and correlations .
in other cases , the parametric family serves as a proxy for a more complex distribution we can not directly sample from .
suppose we have two functions @xmath3 and @xmath4 and we want to compute the expected value @xmath5 with @xmath6 .
if @xmath7 is too large for enumeration of the state space we have to rely on monte carlo algorithms , the vast majority of which involve sampling markov transitions with invariant measure @xmath8 , the standard approach being the metropolis - hastings kernel ( * ? ? ?
* ch . 7 ) . for a transition from @xmath9
, we sample @xmath10 from an auxiliary kernel @xmath1 and accept the step to @xmath11 with probability @xmath12/[\tilde\pi(\v x){\centering{\colorbox{yellow}{\parbox{0.9\textwidth}{(}}}\flushleft{}}\v\gamma\mid \v x)]}\ ] ] or return @xmath13 otherwise .
random walks on @xmath2 are easy to implement but often suffer from slow mixing ; independent proposals @xmath14 provide fast - mixing if @xmath15 is reasonably high on average , in other words if @xmath1 is sufficiently close to @xmath8 @xcite .
this rationale complements other approaches to fast mixing such as parallel chains ( * ? ? ? * among others ) or self - avoiding dynamics @xcite .
+ the vast field of potential applications in monte carlo algorithms encourages the study of families with @xmath16 parameters which , like the multivariate normal , accommodate all valid combinations of means and correlations .
this paper elaborates some theoretical background on random binary vectors , proves the range of possible correlations for a particular class of parametric families , connects to existing work in the literature and provides broad numerical insight concerning the range of dependencies achievable in practice .
it is structured as follows . in section [ sec : properties ] , we introduce suitable notation and review results relating binary distributions to its moments .
section [ sec : families from glms ] elaborates on parametric families which have , by definition , conditional distributions that are generalized linear regressions .
we show that they accommodate the whole range of possible correlations .
section [ sec : log cond family ] motivates the use of the logistic link function as an approximation to the ising - type exponential quadratic family . in section [ sec : sampling ] , we discuss how to adjust the parametric families to specified marginals .
finally , in section [ sec : numerical experiments ] we perform numerical experiments to compare competing approaches for sampling correlated binary data in high dimensions .
we write @xmath0 for the binary space and denote by @xmath17 the generic dimension . given a vector @xmath18 and an index set @xmath19 , we write @xmath20 for the sub - vector indexed by @xmath21 and @xmath22 for its complement . for @xmath23 we use the more explicit notation @xmath24 .
unless otherwise defined , @xmath8 denotes an arbitrary probability mass function on @xmath2 .
we denote by @xmath25 the expected value with respect to @xmath26 and write @xmath27 for an event @xmath28 .
let @xmath29 be a mean vector .
we call @xmath30 the product family or the mass function of @xmath7 independent bernoulli variables . for a set @xmath31
, we refer to @xmath32 as the cross - moment indexed by @xmath21 .
note that @xmath33 which means that cross - moments and marginal probabilities indexed by @xmath31 are identical .
higher order cross - moments coincide with first order cross - moments .
the range of possible cross - moments is limited by the following constraints .
[ prop : bin bounds ] the cross - moments of binary data fulfill the sharp inequalities @xmath34 the lower bound follows from @xmath35 the upper bound is the monotonicity of the measure .
for the special case @xmath36 , proposition [ eq : bin bounds ] is a well - known result and has been invoked in several articles dealing with correlated binary data . for the general case , we remark that a mapping @xmath37^{\abs{i}}\to[0,1],\ f_{i}(m_{i_1},\dots , m_{i_{\abs{i}}})=m_{i}$ ] , which assigns a cross - moment @xmath38 for @xmath31 as function of the marginals @xmath39 for @xmath40 , is quite similar to a @xmath41-dimensional copula and the inequalities are exactly the frchet - hoeffding bounds ( * ? ? ?
[ def : cross - moment matrix ] we say a @xmath42 symmetric matrix @xmath43 with entries in @xmath44 is a _ cross - moment matrix of binary data _ if @xmath45 is positive definite and condition holds for all @xmath31 with @xmath36 .
we derive the family of distributions which , under the constraints that @xmath8 has given cross - moments , maximizes the entropy @xmath46 $ ] .
the following proposition is just a special case of a more general concept @xcite .
[ prop : maxent ] let @xmath47 be a family of index sets such that @xmath48 is a valid set of cross - moments .
the maximum entropy distribution having the specified cross - moments has the form @xmath49 $ ] .
define the lagrange multipliers @xmath50 $ ] and differentiate @xmath51/\partial \pi(\v\gamma)=-\log[\pi(\v\gamma)]-1+\sum_{i\in \mathcal i}a_{i}\prod_{i\in i}\gamma_i$ ] .
solving the first order condition and normalizing completes the proof . for a set @xmath31 ,
we define @xmath52^{-1/2 } $ ] and refer to @xmath53 as the ( generalized ) correlation coefficient indexed by @xmath21 .
a @xmath42 positive definite matrix @xmath54 with entries in @xmath55 $ ] and @xmath56 is not the correlation matrix of a binary distribution for every mean vector @xmath29 .
in fact , @xmath54 is a correlation matrix if and only if @xmath57 is valid in the sense of definition [ def : cross - moment matrix ] , where the dot means point - wise multiplication and @xmath58 .
@xcite elaborate alternative conditions for compatibility between correlations and means , but these do not seem easier to express or to check . in the context of binary data ,
the notion of `` strong correlations '' refers to correlation coefficients which are at the boundary of the feasible range with respect to the mean vector .
note that the absolute value of the correlation coefficient does , in itself , not tell whether the correlation is easy or difficult to model .
the following statement relates the notions of uncorrelated and independent variables .
let @xmath13 be a @xmath7-dimensional binary random vector . for @xmath59 ,
entries are uncorrelated if and only if they are independent . for @xmath60
, entries might be mutually uncorrelated but not independent .
let @xmath61 . by definition @xmath62 .
further , we obtain @xmath63 and , analogously , @xmath64 . finally , we have @xmath65 . for @xmath66 ,
let for instance @xmath67 and @xmath68 .
the entries are mutually uncorrelated , but not independent since @xmath69 .
the following representation by @xcite allows to write a binary distribution in terms of its generalized correlation coefficients . [
prop : bahadur ] let @xmath8 be a binary distribution with mean @xmath29 . then ,
@xmath70.\ ] ] we give the proof by @xcite using the notation introduced above .
the set @xmath71 forms an orthonormal basis on @xmath72 with respect to the inner product @xmath73 therefore , every function @xmath74 has a unique representation @xmath75 .
compute the inner products @xmath76u_i^{\pi}(\v\gamma){\centering{\colorbox{yellow}{\parbox{0.9\textwidth}{^}}}\flushleft{}}\prod_{\v m}(\v\gamma ) = { \mathbb{e}_{\pi}\left(u_i^{\pi}(\v \gamma ) \right)}=c_i^{\pi } $ ] to obtain the desired form @xmath77 .
using proposition [ prop : bahadur ] , we may bound the @xmath78 distance between two binary distribution with the same mean in terms of nearness of their correlation coefficients .
let @xmath8 and @xmath79 be binary distributions with mean @xmath29 .
for @xmath80 , @xmath81 where @xmath82 .
since @xmath83 for all @xmath31 , applying proposition [ prop : bahadur ] yields @xmath84 using that @xmath85 for all @xmath86 , we obtain the bound @xmath87^{1/2}[m_i^{p-1}+(1-m_i)^{p-1 } ] \leq 2^{(1-\min\set{p,2})\abs{i}}.\ ] ] finally , we have @xmath88 , since by definition @xmath89 for all @xmath31 with @xmath90 . [
corr : cross moments ] let @xmath8 and @xmath1 be binary distributions with cross - moment matrix @xmath91 .
then we have @xmath92 . with regard to the metropolis - hastings kernel mentioned in the introductory section , the factor @xmath93 in corollary [ corr :
cross moments ] is the gain of a more complex proposal distribution @xmath94 with @xmath95 over a simple product model @xmath96 with @xmath97 .
the following result shows how the cross - moments of the proposal distribution affect the auto - covariance of the independent metropolis - hastings sampler .
let @xmath8 and @xmath1 be binary distributions with mean @xmath29 and denote by @xmath98 $ ] the metropolis - hastings kernel with invariant measure @xmath8 and proposal distribution @xmath1 where @xmath99 is defined in .
the auto - covariance between @xmath100 and @xmath101 is @xmath102 with @xmath103 where @xmath104 .
we plug the definition of the kernel into the expected value and obtain @xmath105\pi(\v x ) \\ & = m_{ij}^{\pi}+\sum_{\v \gamma,\,\v x\in\b^{d } } ( \gamma_ix_j - x_ix_j ) { \centering{\colorbox{yellow}{\parbox{0.9\textwidth}{(}}}\flushleft{}}\v\gamma)\pi(\v x)\lambda_{{\centering{\colorbox{yellow}{\parbox{0.9\textwidth}{}}}}\flushleft{}}(\v\gamma,\v x ) \\ & = m_im_j+\frac{1}{2}(m_{ij}^{\pi}-m_{ij}^{{\centering{\colorbox{yellow}{\parbox{0.9\textwidth}{}}}}\flushleft{}})+\frac{1}{2}\sum_{\v \gamma,\,\v x\in\b^{d } } ( \gamma_ix_j - x_ix_j ) \abs{{\centering{\colorbox{yellow}{\parbox{0.9\textwidth}{(}}}\flushleft{}}\v\gamma)\pi(\v
x)-{\centering{\colorbox{yellow}{\parbox{0.9\textwidth}{(}}}\flushleft{}}\v x)\pi(\v\gamma)},\end{aligned}\ ] ] where we used @xmath106 .
the triangle inequality @xmath107 = 2\sum_{\v \gamma\in\b^{d}}\abs{\pi(\v\gamma)-{\centering{\colorbox{yellow}{\parbox{0.9\textwidth}{(}}}\flushleft{}}\v\gamma)}.\end{aligned}\ ] ] yields the bound on @xmath108 . for some applications , it suffices to model structured dependencies , such as exchangeable @xmath109 , moving average @xmath110 or autoregressive @xmath111 correlations for @xmath112 .
there is a long series of articles concerned with efficient approaches to sampling binary vectors for structured correlations @xcite . in this paper
, we focus on the problem of sampling binary data with arbitrary cross - moment matrix .
we want to construct a parametric family @xmath1 for sampling independent random vectors with specified mean and correlations . sampling in high dimensions , however , requires the computation of conditional distributions @xmath113 , and it is therefore convenient to define the parametric family directly in terms of its conditionals .
let @xmath114 $ ] be a monotonic function and @xmath115 a @xmath42 real - valued lower triangular matrix .
we refer to @xmath116^{\gamma_i}\left[1-\link(a_{ii}+\sum_{j=1}^{i-1}a_{ij}\gamma_j)\right]^{1-\gamma_i},\end{aligned}\ ] ] as the @xmath117-conditionals family .
[ cor : mu product ] let @xmath114 $ ] be a monotonic bijection and @xmath118 a mean vector . for @xmath119 $ ] we have @xmath120 . by construction
, it is straightforward to sample @xmath121 and evaluate @xmath122 point - wise as summarized in procedure [ algo : sampling ] .
alternatively , one could sample from an auxiliary distribution @xmath123 on @xmath124 which allows to compute @xmath125 and define a parametric family @xmath126 through the mapping @xmath127 .
we come back to this idea in section [ sec : gaussian ] .
@xmath128 @xmath129 , @xmath130}$ ] + * if @xmath131 * then @xmath132 + @xmath133 + @xmath134 * * [ algo : sampling ] @xcite discusses the @xmath135-conditionals family with a truncated linear link function @xmath136 .
the linear structure allows to compute the parameters by simple matrix inversion ; on the downside , the linear function is truncated and fails to accommodate complicated correlation structures .
therefore , @xcite elaborates on conditions that guarantee the linear conditionals family to be valid for special correlation structures .
@xcite propose a @xmath135-conditionals family with a logistic link function @xmath137 $ ] .
however , they only analyze the special case of autoregressive correlation structure . in section
[ sec : log cond family ] , we further motivate the use of the logistic link function which indeed allows to model any feasible correlation structure as states the following theorem .
[ thm : mean+corr ] let @xmath138 $ ] be a monotonic , differentiable bijection and @xmath91 a @xmath42 cross - moment matrix .
there is a unique @xmath42 real - valued lower triangular matrix @xmath139 such that @xmath140 . besides the logistic function invoked above ,
popular link functions include the complementrary log - log function with @xmath141 $ ] and the probit function with @xmath142 @xcite .
we derive two auxiliary results to structure the proof of theorem [ thm : mean+corr ] .
[ lem : ext cross - moments ] for a cross - moment matrix @xmath91 with mean vector @xmath143 , we have @xmath144 note that @xmath145 because the covariance matrix @xmath146 is positive definite .
dividing by @xmath147 we obtain @xmath148 which yields @xmath149 = \det(\m m ) \det \begin{pmatrix } \m i & \m m^{-1}\v m \\
0\t & ( 1-\v m\t\m m^{-1}\v m ) \end{pmatrix } \\[1ex ] & = \det(\m m)(1-\v m\t\m m^{-1}\v m)>0.\end{aligned}\ ] ] therefore , all principal minors are positive .
[ lem : bijective ] let @xmath138 $ ] be a monotonic , differentiable bijection , and denote by @xmath150 the open ball with radius @xmath151 .
let @xmath8 be a binary distribution with cross - moment matrix @xmath91 .
we write @xmath143 and @xmath152 for the mean vector .
there is @xmath153 such that the function @xmath154 is a differentiable bijection .
we set @xmath155 for @xmath156 , the partial derivatives of @xmath157 are @xmath158 we have @xmath159 since @xmath117 is strictly monotonic .
then the jacobian is positive for all @xmath160 , @xmath161 \geq \eta_r^{d+1}\ , \det \begin{pmatrix } \m m & \v m \\
\v m\t & 1 \end{pmatrix } > 0,\end{aligned}\ ] ] where we applied lemma [ lem : ext cross - moments ] in the last inequality .
we proceed by induction over @xmath7 . for @xmath162 , @xmath163 is a scalar and we define the @xmath135-conditionals family @xmath164 via corollary [ cor : mu product ] .
suppose that we have already constructed a @xmath135-conditionals family @xmath165 with @xmath42 lower triangular matrix @xmath166 and cross - moment matrix @xmath167 .
we can add a new dimension to the @xmath135-conditionals model @xmath165 without changing @xmath167 , since @xmath168^{\xi_{d+1}}\times\\ & \qquad \left[1-\link(a_{d+1,d+1}+\textstyle\sum_{j=1}^{d } a_{d+1,j}\xi_j)\right]^{1-\xi_{d+1 } } \\ = & \ \sum_{\v\gamma\in\b^d } { \centering{\colorbox{yellow}{\parbox{0.9\textwidth}{_}}}\flushleft{}}{\m a(d)}^{\link}(\v\gamma ) \bigg\ { \link(a_{d+1,d+1}+\textstyle\sum_{j=1}^{d } a_{d+1,j}\gamma_j ) \begin{pmatrix } \v\gamma\v\gamma\t & \v\gamma \\ \v\gamma\t & 1 \end{pmatrix}+\\ & \qquad \left[1-\link(a_{d+1,d+1}+\textstyle\sum_{j=1}^{d } a_{d+1,j}\gamma_j)\right ] \begin{pmatrix } \v\gamma\v\gamma\t & \v0 \\ \v0\t & 0 \end{pmatrix } \bigg\ } \\ = & \ \sum_{\v\gamma\in\b^d } { \centering{\colorbox{yellow}{\parbox{0.9\textwidth}{^}}}\flushleft{}}{\link}_{\m a(d)}(\v\gamma ) \link(a_{d+1,d+1}+\textstyle\sum_{j=1}^{d } a_{d+1,j}\gamma_j ) \begin{pmatrix } \m 0 & \v\gamma \\ \v\gamma\t & 1 \end{pmatrix}+\\ & \qquad \begin{pmatrix } \m m(d ) & \v0 \\
\v0 \t & 0 \end{pmatrix}\end{aligned}\ ] ] for reasons of symmetry , it suffices to show that there is @xmath169 such that @xmath170 where the r.h.s . denotes the @xmath171th column of the augmented cross - moment matrix .
there is @xmath172 so that @xmath173 with @xmath174\t,1)$ ] which implies that a solution is contained in a sufficiently large open ball @xmath175 .
we apply lemma [ lem : bijective ] to complete the inductive step and the proof .
we denote by @xmath176 the logistic conditionals family , that is the @xmath135-conditionals family with logistic link function @xmath177 $ ] .
this parametric family has been proposed by @xcite , and in more general terms suggested by @xcite . in this section ,
we motivate why the logistic link function arises somewhat naturally in the context of @xmath135-conditional families .
let @xmath139 be a @xmath42 real - valued lower triangular matrix .
we refer to @xmath178 as the exponential quadratic family with @xmath179 $ ] . if @xmath180 , then @xmath181 and @xmath182 .
the exponential quadratic family is a natural way to design a parametric family and plays a central role in physics and life science being the well - studied ising model on a weighted complete graph .
it links to information theory @xcite , log - linear theory for contingency tables ( * ? ? ?
* ch . 5 ) and graphical models ( * ? ? ?
finding its mode is an np - hard problem and intensively studied in the field of operation research ( * ? ? ?
* for a recent review ) .
proposition [ prop : maxent ] states that the exponential quadratic family is the maximum entropy distribution on @xmath183 having a given cross - moment matrix .
it appears to be the binary analogue of the multivariate normal distribution which is the maximum entropy distribution on @xmath184 having a given covariance matrix @xcite .
we can read the parameters @xmath185 as lagrange multipliers or , if @xmath186 , as conditional log odd - ratios since @xmath187.\ ] ] we might interpret the constant conditional log odd - ratios as analogue of the constant conditional correlations of the multivariate normal distribution @xcite .
despite these similarities to the multivariate normal distribution , we can not easily sample from the exponential quadratic family nor explicitly relate the parameter @xmath139 to the cross - moment matrix @xmath91 .
the reason is that the lower dimensional marginal distributions are difficult to compute ( * ? ? ?
* ( iii ) ) .
the marginal distribution of the exponential quadratic family is @xmath188\right).\ ] ] we can not repeat the marginalization since the multi - linear structure is lost . in fact , the following result shows that the logistic conditionals family is precisely constructed such that the non - linear term in the above expression vanishes .
let @xmath139 be a @xmath42 lower triangular matrix .
the logistic conditionals family can be written as @xmath189\right).\end{aligned}\ ] ] straightforward calculations yield @xmath190^{\gamma_i}[1-\logistic(a_{ii}+\sum_{j=1}^{i-1}a_{ij}\gamma_j)]^{1-\gamma_i}\right ) \\ & \textstyle= \sum_{i=1}^d\left(\gamma_i\log[\logistic(a_{ii}+\sum_{j=1}^{i-1}a_{ij}\gamma_j)]+ ( 1-\gamma_i)\log[1-\logistic(a_{ii}+\sum_{j=1}^{i-1}a_{ij}\gamma_j)]\right ) \\ & \textstyle= \sum_{i=1}^d\left(\gamma_i\,\logistic^{-1}[\logistic(a_{ii}+\sum_{j=1}^{i-1}a_{ij}\gamma_j)]+ \log[1-\logistic(a_{ii}+\sum_{j=1}^{i-1}a_{ij}\gamma_j)]\right ) \\ & \textstyle= \sum_{i=1}^d\left(\gamma_i(a_{ii}+\sum_{j=1}^{i-1}a_{ij}\gamma_j)- \log[1+\exp(a_{ii}+\sum_{j=1}^{i-1}a_{ij}\gamma_j)]\right ) \\ & \textstyle= \sum_{i=1}^d\sum_{j=1}^i
a_{ij}\gamma_i\gamma_j-\sum_{i=1}^d \log[1+\exp(a_{ii}+\sum_{j=1}^{i-1 } a_{ij}\gamma_j ) ] \\ & \textstyle= \v\gamma\t\m
a\v\gamma-\sum_{i=1}^d \log[1+\exp(a_{ii}+\sum_{j=1}^{i-1 } a_{ij}\gamma_j)],\end{aligned}\ ] ] where we used @xmath191=-\log[1+\exp(x)]$ ] in the third line .
the full conditional probability of the @xmath7-dimensional exponential quadratic family is a logistic regression term .
the conditional distribution of the exponential quadratic family is @xmath192 since we can not repeat the marginalization for lower dimensions , we can not assess the lower dimensional conditional probabilities which are necessary for sampling .
we can , however , derive a series of approximate marginal probabilities that produce a logistic conditionals family which is , for low correlations , close to the original exponential quadratic family .
this idea goes back to @xcite .
let @xmath193 $ ] be a second order approximation .
we may approximate the marginal distribution @xmath194 by an exponential quadratic family @xmath195 with parameters @xmath196 where @xmath197 denotes the @xmath7th column of @xmath139 without @xmath198 .
we write the marginal distribution of the exponential quadratic family as @xmath199\right)\big].\end{aligned}\ ] ] using the identity @xmath200 = \textstyle\log\big(\exp(\frac{1}{2}x)\big[\exp(-\frac{1}{2}x)+\exp(\frac{1}{2}x)\big]\big ) = \textstyle\frac{1}{2}x+\log\big[2\cosh(\frac{1}{2}x)\big]\ ] ] and approximate the non - quadratic term by the second order polynomial @xmath201\approx c_1+c_2 \v a_{*}\t\v\gamma_{-d}+c_3 ( \v a_{*}\t\v\gamma_{-d})^2.\ ] ] we rewrite the inner products @xmath202\v\gamma_{-d}$ ] and rearrange the quadratic terms .
we can iterate the procedure to construct a logistic conditionals family which is close to the original exponential quadratic family .
however , the function @xmath203 $ ] behaves like a quadratic function around zero and like the absolute value function for large @xmath204 .
thus , a quadratic polynomial can only approximate @xmath203 $ ] well for small values of @xmath205 which means that exponential quadratic families with strong dependencies is hard to approximate .
@xcite propose a taylor approximation which fits well around @xmath206 and works for weak correlations .
the parameters are @xmath207),\frac{1}{2}\tanh(\frac{1}{2}a_{dd}),\frac{1}{8}\,\mathrm{sech}^2(\frac{1}{2}a_{dd})\big)$ ] .
if @xmath208 full probabilities are known , we easily sample from the corresponding multinomial distribution @xcite . for a valid set of cross - moments @xmath209 , @xmath210
, @xcite proposes to compute the full probabilities using a variant of the iterative proportional fitting algorithm @xcite .
while there are no restrictions on the range of dependencies , we have to enumerate the entire state space which limits this versatile approach to low dimensions . in the sequel
, we do not consider methods for structured correlations nor approaches which require enumeration of the state space .
first , we show how to compute the parameter @xmath139 of a @xmath135-conditionals model for a given cross - moment matrix @xmath91 .
secondly , we review an alternative approach to sampling binary data based on the multivariate normal distribution @xcite .
the proof of theorem [ thm : mean+corr ] suggests an iterative procedure to adjust the parameter @xmath139 to a given cross - moment matrix @xmath91 .
we add new cross - moments @xmath211 to the @xmath42 a lower triangular matrix @xmath139 by solving the non - linear equation @xmath212 via newton - raphson iterations @xmath213^{-1}[f(\v a^{(k)})-\v m]$ ] where @xmath214(\v\gamma\t,1)\t \\ f'(\v a)&=\textstyle\sum_{\v\gamma\in\b^{d } }
{ \centering{\colorbox{yellow}{\parbox{0.9\textwidth}{_}}}\flushleft{}}{\m a}^{\link}(\v\gamma)\link'[(\v\gamma\t,1)\v a](\v\gamma\t,1)\t(\v\gamma\t,1)\end{aligned}\ ] ] for dimensions @xmath215 , the exact computation of the expectations becomes expensive , and we replace @xmath157 and @xmath216 by their monte carlo estimates @xmath217(\v x_k\t,1 ) \\
\hat f'(\v a)&=\textstyle\sum_{k=1}^n { \centering{\colorbox{yellow}{\parbox{0.9\textwidth}{_}}}\flushleft{}}{\m a}^{\link}(\v\gamma)\link'[(\v x_k\t,1)\v a)](\v x_k\t,1)\t(\v x_k\t,1 ) \end{aligned}\ ] ] where @xmath218 are drawn from @xmath219 .
some remarks are in order . *
if the smallest eigenvalue of @xmath220 approaches zero or a cross - moment @xmath221 approaches the bounds , the parameter @xmath185 may become very large in absolute value .
the limited numerical accuracy available on a computer inhibits sampling from such extreme cases .
* we might encounter numerical trouble in the course of the fitting procedure . in order to circumvent problems
, we set @xmath222 for all @xmath223 and compute a sequence of solutions @xmath224 to the cross - moments @xmath225 .
we stop if the parameters fail to converge which ensures that the mean of the @xmath135-conditionals family is always @xmath226 . *
if we have data available instead of cross - moments , we would rather fit the family via component - wise likelihood maximization which is usually faster than the method of moments and can even be parallelized @xcite .
* for the linear link function @xmath227 , we obtain @xmath228\v a = \begin{pmatrix } \m m & \v m \\ \v m\t & 1 \end{pmatrix}\v a\end{aligned}\ ] ] which always has a solution by virtue of lemma [ lem : ext cross - moments ] ; to construct a mass function , however , we have to fall back to the truncated version @xmath136 , and the range of feasible cross - moments is hard to assess @xcite .
@xcite propose to dichotomize a multivariate gaussian distribution for sampling multivariate binary data . for a vector @xmath229 and a @xmath42 correlation matrix @xmath230
we define the gaussian copula family @xmath231 where @xmath232}(x_1),\dots,\ind_{(-\infty , a_d]}(x_d)\right)$ ] . for all @xmath31 , the marginals are @xmath233 \scriptscriptstyle ( -\infty , a_i ] & \scriptstyle \hspace{-.7em } i\in i \\[-.5em ] \scriptscriptstyle ( -\infty,\infty ) & \scriptstyle \hspace{-.7em } i\notin i \end{cases}}\ , \varphi_{\g \sigma}(\v v)\,d\v v=\phi_{\g \sigma}^{(i)}(\v a_i),\end{aligned}\ ] ] where @xmath234 is the marginal cumulative distribution function of the multivariate gaussian .
we set @xmath235 for @xmath236 to adjust the mean . in order to compute the parameter @xmath230 that yields the desired cross - moments @xmath91
, we may use a fast series approximations @xcite to solve @xmath237 for @xmath238 via newton - raphson iterations @xmath239/\varphi_{\sigma_{ij}^r}(a_i , a_j)$ ] ; @xcite suggests the bivariate @xcite distribution as a proxy for @xmath240 which might provide a good starting value @xmath241 . while we always obtain a solution in the bivariate case , it is well - known that the resulting matrix @xmath230 is not necessarily positive definite due to the range of the gaussian copula which allows to attain the bounds for @xmath242 , but not for higher dimensions . in that case , we can replace @xmath230 by @xmath243 where @xmath244 is smaller than any eigenvalue of @xmath230 .
alternatively , we can project @xmath230 into the set of correlation matrices ; see @xcite and follow - up papers for algorithms that compute the nearest correlation matrix in frobenius norm .
the point - wise evaluation of @xmath245 requires the computation of multivariate normal probabilities , that is high - dimensional integrals with the respect to the density of the multivariate normal distribution .
this is a computationally challenging task in itself ( see e.g. * ? ? ?
* ) , and the gaussian copula family is therefore not easily incorporated into the markov chain monte carlo algorithms briefly discussed in the introduction .
in this section , we compare the @xmath135-conditionals family with truncated linear and logistic link function to the gaussian copula family .
we draw random cross - moment matrices of varying dimension and difficulty , fit the parametric families and record how well the desired correlation structure can be reproduced on average .
we first sample the mean @xmath246 .
for the off - diagonal elements , we have to ensure that the covariance matrix @xmath146 is positive definite and that the constraints are all met .
we alternate the following two steps . * permutations @xmath247 for @xmath248 with uniform @xmath249 where we denote by @xmath250 the set of all permutations on @xmath251 .
* replacements @xmath252}$ ] for all @xmath253 with uniform @xmath254 where the bounds @xmath255 are subject to the constraints @xmath256 and @xmath257 .
the replacement step needs some consideration .
we denote by @xmath258 the inverse of the @xmath259 upper sub - matrix of @xmath91 and define @xmath260 such that @xmath261 ( m_{dd}-\sum_{i\in d\setminus\set d}\tau_i).$ ] if we replace @xmath262 by @xmath263 we have to ensure that @xmath264=\det(\m m)+m_{di}(m_{di}n_{ii}+2\tau_{i})-x_i(x_in_{ii}+2\tau_{i})>0 $ ] which means @xmath265 with @xmath266^{-1/2}$ ] . therefore , the lower and upper bounds , @xmath267 and @xmath268 , respect all constraints on @xmath263 .
we rapidly update the value of the determinant @xmath264 $ ] and proceed with the next entry .
we perform @xmath269 permutation steps and run @xmath270 sweeps of replacements between permutations .
the result is approximately a uniform draw from the set of feasible cross - moments matrices .
however , sampling according to these cross - moments might not be possible in higher dimensions because the cross - moment matrix is likely to contain extreme cases which are beyond the scope of the parametric family or not workable for numerical reasons .
we introduce a parameter @xmath271 $ ] which governs the difficulty of the sampling problem by shrinking the upper and lower bounds @xmath272 and @xmath273 of the uniform distributions to @xmath274/2 $ ] and @xmath275/2 $ ] , respectively .
let @xmath91 be a cross - moments matrix and let @xmath276 denote the cross - moment matrix with mean @xmath143 and uncorrelated entries @xmath277 for all @xmath112 . for a parametric family @xmath278 , we define the figure of merit @xmath279 where @xmath280 denotes the sampling cross - moment matrix of the parametric family with parameter @xmath281 adjusted to the desired cross - moment matrix @xmath91 .
the norm @xmath282 might be any non - trivial matrix norm ; in our numerical experiments we use the spectral norm @xmath283 , where @xmath284 delivers the largest eigenvalue , but we found the frobenius norm @xmath285 to provide qualitatively the very same picture . for fitting the logistic conditionals family when @xmath215 , we replace the exact terms by monte carlo estimates where we use @xmath286 random samples .
we estimate the cross - moment matrix of the parametric family @xmath1 by @xmath287 where we use @xmath288 samples from @xmath1 .
this concerns only the logistic and linear conditionals families ; for the gaussian copula family , we can explicitly compute the sampling cross - moments as @xmath289 , where @xmath230 is the adjusted correlation matrix of the underlying multivariate normal distribution made feasible via .
we loop over @xmath290 levels of difficulty @xmath271 $ ] in @xmath291 dimensions @xmath292 , and generate at each time @xmath293 cross - moments matrices .
we denote by @xmath294 the ordered figures of merit of the random cross - moment matrices .
we report the median and the quantiles @xmath295 , depicted as underlying gray areas for @xmath296 equidistant values of @xmath297 $ ] .
figures 1 - 3 show the results grouped by parametric families ; the @xmath298-axis with the scale on the left represents the figure of merit @xmath299 $ ] , the @xmath205-axis represents the level of difficulty @xmath271 $ ] , and the @xmath300$]-gray - scale on the right refers to the level of the quantiles .
+ @xmath301@xmath302@xmath303 + @xmath301@xmath302@xmath303 + + @xmath301@xmath302@xmath303 + while theorem [ thm : mean+corr ] suggests that the scope of the logistic conditionals family is far beyond competing approaches , we can not , in practice , expect a binary parametric family with @xmath304 dependency parameters to produce just any desired correlation structure .
however , the practical scope of the logistic family is limited only by the available numerical accuracy while the scope of competing methods is also limited by their mathematical structure .
the truncated linear conditionals family is fast to compute but its quality deteriorates rapidly with growing complexity .
the gaussian copula family is guaranteed to have the correct mean but it is less flexible than the logistic conditionals family ; besides , it does not allow for point - wise evaluation of its mass function .
the logistic conditionals family is computationally demanding but by far the most versatile option .
these findings confirm similar comparisons carried out against the backdrop of particular applications @xcite .
this work is part of the author s ph.d .
thesis at crest under supervision of nicolas chopin whom i would like to thank for numerous discussions on this topic .
i thank ioannis kosmidis for his comments on a prior version of this paper . | we discuss a class of binary parametric families with conditional probabilities taking the form of generalized linear models and show that this approach allows to model high - dimensional random binary vectors with arbitrary mean and correlation .
we derive the special case of logistic conditionals as an approximation to the ising - type exponential distribution and provide empirical evidence that this parametric family indeed outperforms competing approaches in terms of feasible correlations . |
in studies of the cosmic large - scale structure , a number of different statistical methods are routinely used to extract various information of interest ( e.g. , regarding the cosmology , the initial perturbation , etc ) that is embedded in the distribution of the dark matter particles ( in the case of simulations ) or the galaxies ( in observations ) .
the power spectrum @xmath1 is one of the most powerful and basic statistical measures that describes the distribution of mass and light in the universe , and one of the most throughly investigated quantities in modelling the structure formation process .
the initial primordial power spectrum of the mass fluctuations is usually assumed to follow a power law , @xmath5 .
the linearly processed power spectrum @xmath6 can be well predicted by codes such as cmbfast ( seljak & zaldarriaga 1996 ) , or approximated by various fitting formula ( e.g. bardeen et al .
1986 ; efstathiou , bond & white 1992 ; eisenstein & hu 1998 ) for different matter and energy content .
using n - body simulations , the non - linear power spectrum @xmath7 has been modelled by various authors ( e.g. peacock & dodds 1996 ; ma & fry 2000 ; smith et al . 2003 ) .
apart from these theoretical models , direct measurement of the power spectrum from observations plays an extremely important role both in cosmology and galaxy formation theories .
although there are different biases relative to the mass power spectrum , one can roughly say that @xmath1 on very large scales measures the primordial density fluctuations , which is closely connected with the cosmology models ( e.g. , spergel et al .
2007 ) , while @xmath1 on small scales characterizes the later non - linear evolution ( e.g. , peacock & dodds 1996 ) . as an essential statistical measure for the distribution of galaxies ,
the power spectrum @xmath1 has been estimated and modeled from most of the redshift surveys .
recent investigations along this direction include the cfa and perseus - pisces redshift surveys ( baumgart & fry 1991 ) , the radio galaxy survey ( peacock & nicholson 1991 ) , the iras qdot survey ( kaiser 1991 ) , the 2jy iras survey ( jing & valdarnini 1993 ) , the 1.2jy iras survey ( fisher et al .
1993 ) , the las campanas redshift survey ( lin et al .
1996 ; yang et al . 2001 ) , the 2df galaxy redshift survey ( percival et al . , 2001 ; tegmark , hamilton & xu 2002 ; snchez et al . 2006 ) and the sloan digital sky survey ( tegmark et al . 2004 ; percival et al . , 2007 ) .
among these works , the galaxy power spectra are measured either using the fast fourier transform ( fft ) technique or direct summation , or other advanced techniques ( e.g. , yang et al .
2001 ; tegmark et al .
2004 ) . apart from these observational probes , the power spectrum is also widely measured from n - body simulations ( e.g. davis et al .
1985 ) . for these measurements ,
one has to use ffts since there are too many particles in the simulations to apply direct summation . before performing the fft ,
one therefore needs to assign the particle distribution @xmath8 onto grids @xmath9 ( usually onto @xmath10 grid cells , where @xmath11 is an integer ) . as pointed out in a recent paper by jing ( 2005 ) ,
such an assignment process is equivalent to a convolution of the real density field with a given assignment window function @xmath12 , and sampling the convolved density field at the @xmath10 grid points .
thus the power spectrum based on the fft of @xmath13 is not equal to that based on the fourier transform ( ft ) of @xmath8 . in order to obtain the true power spectrum to an accuracy of a few percent ,
the sampling effects should be carefully corrected ( jing 2005 ; and references therein ) . to this end , jing ( 2005 ) proposed an elegant algorithm to iteratively deconvolve the power spectrum measurement for the impact of the mass assignment and to extract the true power spectrum .
tests using n - body simulations show that their algorithm works extremely well for the three commonly used mass assignment functions , i.e. , the nearest grid point ( ngp ) , the cloud in cell ( cic ) and the triangular shaped cloud ( tsc ) methods . in this paper , rather than trying to correct for the influence of the window function , we seek to minimize the effects of the mass assignment by selecting special window functions .
an ideal window function should fulfill the following criteria : ( i ) compact top - hat like support in fourier space to avoid the sampling effects ; ( ii ) compact support in real space to allow computationally efficient mass assignment onto grids .
we find that the scale functions of the daubechies wavelet transformations are good candidates for simultaneously matching both requirements .
in fact , as we will demonstrate they allow an accurate measurement of the power spectrum with ffts without the need for a deconvolution procedure .
this is of great help especially for accurate measurements of higher order spectra , like the bi - spectrum , where ffts are needed but the de - aliasing methods are not available yet .
we will discuss this application to accurate measurements and modelling of the bi - spectrum in a subsequent paper .
this paper is organized as follows . in section [ sec_form ]
we give a brief description of the methodology for measuring the power spectrum from the discrete distribution of dark matter particles . in section [ sec_window ]
we first present the commonly used window functions in both real and fourier spaces , and then and introduce our new mass assignment scheme . in section [ sec_test ]
we compare the power spectra extracted from the millennium run using different methods .
finally , we summarize our results in section [ sec_summ ] .
in this section we outline the methods used to measure the power spectrum from the distribution of dark matter particles ( peebles 1980 ) . unless stated otherwise we shall follow jing ( 2005 ) , and we refer readers who are interested in a more detailed and complete set of formulae to this paper and references therein .
we start from the definition of the power spectrum .
let @xmath14 be the cosmic density field and @xmath15 the mean density .
then the density contrast @xmath16 can be expressed as , @xmath17 based on the cosmological principle , we assume that @xmath18 in a very large volume @xmath19 fairly represents the overall cosmic density field , and that it can be taken to be periodic .
the ft of @xmath20 can be defined as : @xmath21 and by definition , its power spectrum @xmath1 is simply related to @xmath22 as @xmath23 where @xmath24 means the ensemble average .
however , in practice , the cosmic density field is usually traced by the distribution of galaxies or dark matter particles . in these cases , the density field @xmath14
is replaced by the number density distribution of objects @xmath25 , where @xmath26 is the coordinate of object @xmath27 and @xmath28 is the dirac-@xmath29 function . and
the ft of the related number density contrast @xmath30 can be expressed as , @xmath31 where @xmath32 is the mean number density , the superscript @xmath33 represents the _ discrete _ case of @xmath14 , and @xmath34 is the kronecker delta .
if we divide the volume @xmath19 into infinitesimal elements @xmath35 within which there are either 0 or 1 objects , then the above equation can be written as : @xmath36 where @xmath37 is the total number of objects in @xmath19 and @xmath38 is either 0 or 1 . after a bit of algebra , it is seen that the true power spectrum can be measured via @xmath39 obviously , when the ft is directly applied to the distribution of the galaxies or dark matter particles , one needs to correct for the discreteness ( or shot noise ) effect , which introduces an additional term @xmath40 to the power spectrum @xmath41 . the above method of using a direct summation in the ft can be used to measure the power spectrum from the distribution of galaxies , when the number of objects is not very large . however , because of the huge number of particles involved in n - body simulations , it is almost impossible to be applied to the dark matter particles of cosmological density fields .
instead , a computationally attractive approach is to use an fft .
the density contrast in fourier space using a fft is , @xmath42 where the superscript @xmath43 represents the fft .
@xmath44 is the convolved density value on the @xmath45-th grid point @xmath46 ( where @xmath45 is an integer vector ; @xmath47 is the grid spacing ) , @xmath48 where @xmath49 is the mass assignment function .
note that eqs .
( [ eq : deltak_d ] ) and ( [ eq : deltak_f ] ) are different in that the summations carried out in the former equation is over the objects and the latter over space ( the grid points ) .
after several steps ( see also hockney and eastwood 1981 ) , jing ( 2005 ) derived the following power spectrum estimator , @xmath50 where @xmath51 is the ft of the window function @xmath49 , @xmath52 is the nyquist wavenumber , and the summation is over all 3d integer vectors @xmath53 . according to equation ( [ eq : p_f ] ) , one can easily identify the impact of the mass assignment onto the measured power spectrum .
first , the mass assignment introduces the factor @xmath54 to both the true power spectrum and the shot noise ( @xmath40 ) terms .
second , the quantity @xmath55 is a measure for a _ convolved _ power spectrum ( i.e. , the sums over @xmath56 ) which suffers from _ sampling _ effects . as pointed out in jing ( 2005 ) and will be shown in section [ sec_window ] , the sampling effects are significant near the nyquist wavenumber @xmath57 and should be carefully corrected in an accurate measure of the power spectrum .
as shown by equation ( [ eq : p_f ] ) , the mass assignment window function plays an important role in measuring the power spectrum using an fft .
we separate its impact into two parts : one on the shot noise term ( second term of the r.h.s . of eq .
[ eq : p_f ] ) and the other on the true power spectrum term ( first term of the r.h.s . of eq .
[ eq : p_f ] ) .
hereafter we refer to the impact on the true power spectrum term as the _ sampling effects_. usually , the impact on the shot noise term can be handled analytical according to the ft of the window function . however , because of the convolution with the true power spectrum , the sampling effects can not be corrected easily .
there are basically two strategies for handling the sampling effects .
one can either try to correct for them by deconvolving the impact of the window function ( which is carried out in jing 2005 ) or try to use an optimal window function that minimizes the sampling effects from the outset ( the purpose of this work ) .
below we discuss a few commonly used window functions as well as the particular mass assignment proposed here both in real and fourier spaces , and then discuss their impact on measuring the true power spectrum with an fft in detail .
the most popular mass assignment functions used in measuring the power spectrum are the ngp , cic and tsc methods .
their forms in real space can be described by @xmath58 , with @xmath59 @xmath60 and @xmath61 where @xmath62 ( @xmath63 ) is the @xmath11-th component of @xmath64 . in the left panel of fig .
[ fig : assign_1 ] , we show these window functions in real space , with solid , dotted and dashed lines corresponding to the ngp , cic and tsc methods , respectively .
their impact on the measurement of the power spectrum using an fft ( eq . [ eq : p_f ] ) can be understood most easily based on their fourier space behavior . according to hockney & eastwood ( 1981 )
, these three mass assignment window functions can be described in fourier space by @xmath65 , with @xmath66^p\,,\ ] ] where @xmath67 ( @xmath63 ) is the @xmath11-th component of @xmath68 , and @xmath69 for ngp , @xmath70 for cic , and @xmath71 for tsc .
we show in the right panel of fig .
[ fig : assign_1 ] ( the square of ) the related window functions in fourier space .
these window functions peak at @xmath72 with @xmath73 and decrease sharply with @xmath74 , especially for the cic and tsc kernels . according to eq .
( [ eq : p_f ] ) , the impact of the window functions can be separated into two parts , one on the shot noise and one on the true power spectrum .
it is quite easy to model and correct the impact on the shot noise term , @xmath75 for the ngp , cic , and tsc assignments , the shot noise term can be expressed as , @xmath76 , & cic , \cr \pi_{i}[1-\sin^2({\pi k_i\over 2k_n})+ { 2\over 15}\sin^4({\pi k_i\over 2k_n } ) ]
. & tsc.\cr}\ ] ] in practice , for the latter two cases , one often uses the following isotropic approximation to model the shot noise term , @xmath77 , & cic , \cr [ 1-\sin^2({\pi k\over 2k_n})+{2\over 15}\sin^4({\pi k\over 2k_n } ) ] , & tsc,\cr}\ ] ] where @xmath78 .
as has been shown in jing ( 2005 ) , this approximation works very well for @xmath79 , however , it can underestimate the true value by about 40% at @xmath80 .
nevertheless , compared to the power spectrum term we are trying to measure in a cdm cosmology , this error in the shot noise term is usually negligible .
now we turn to the impact of the window functions on the first term of the r.h.s of eq .
( [ eq : p_f ] ) , the sampling effects .
there are three aspects that need to be taken into account in measuring the true power spectrum if an accuracy of a few _ percent _ is required .
* smearing effect + in the summation of the true power spectrum over @xmath53 , only the term @xmath81 is what we intend to measure .
however , according to the results shown in the right panel of fig .
[ fig : assign_1 ] , the @xmath82 term decreases sharply from @xmath83 at @xmath74 , especially for the cic and tsc methods . thus the contribution from the _ related _ true power spectrum
@xmath84 is greatly suppressed .
this effect is the so - called smearing or smoothing effect , which has been discussed in the literature ( e.g. , baumgart & fry 1991 ; jing & valdarnini 1993 ; scoccimarro et al .
1998 ) * anisotropy effect + in pratice , one may use the average of the @xmath85 over different directions for a given @xmath4 to estimate the power spectrum @xmath1 .
however , the window function @xmath86 is not isotropic for different directions for a given @xmath4 , that is , @xmath86 is different , e.g. , for @xmath87 , @xmath88 , @xmath89 , etc .
this effect is small for the ngp method , but quite significant for the cic and tsc methods , especially at @xmath80 ( eg .
baumgart & fry 1991 , jing 2005 ) .
* aliasing effect + the power spectrum estimator @xmath55 contains not only the contribution from @xmath84 where @xmath81 but also from @xmath90 where @xmath91 .
the latter contribution , which is called the alias effect , prevents us from obtaining the true power spectrum @xmath92 straightforwardly .
this effect , which is prominent near the nyquist wavenumber @xmath93 ( significant for ngp method and less significant for tsc method ) , has been discussed and handled using an iterative correction method in jing ( 2005 ) .
the smearing and anisotropy effects are easy to be corrected .
for instance , one can directly normalize the density contrast in fourier space , @xmath94 , with the window function @xmath95 ( e.g. , baumgart & fry 1991 ) .
thus , the corrected density contrast @xmath96 obviously no longer suffer from these two effects at @xmath97 , however at the price of a much enhanced aliasing effect ( i.e. , the @xmath98 terms in eq .
[ eq : p_f ] ) .
because of the aliasing effect the power spectrum measured at @xmath99 can become a factor of 2 larger than the true value .
such kind of aliasing effects also exist in radio imaging analyses based on ffts , and various perticular mass assignment schemes have been discussed in order to minimize their impact ( e.g. briggs et al .
1999 ) . using an elegant iterative correction methods , jing ( 2005 )
has properly corrected the impact of the aliasing effect , and illustrated its success in obtaining the true power spectrum . on the other hand , his method
can only be applied to the estimation of the power spectrum . for measurements of higher order spectra ,
e.g. the bi - spectrum , there is so far no straightforward method that can correct the aliasing effect . in what follows , rather than trying to correct the above three kinds of effects , we attemp to find a mass assignment window function that does not or only to very small degree suffer from these effects . an ideal window function that does not suffer from the sampling effects mentioned above is obviously a top - hat in fourier space .
using such a window function , the power spectrum measurement eq .
( [ eq : p_f ] ) can be reduced to eq .
( [ eq : p_d ] ) .
however such a mass assignment function , @xmath100 , is not a compact localized function in real space . in the mass assignment onto the grid
, one may then have to distribute each particle s mass to too many grid cells .
in fact , if we want to maintain an accuracy of 1% in the mass assignment , the mass of each particle should be distributed to @xmath101 grid cells !
such an assignment scheme may eliminate most if not all of the computational advantage that an fft can bring us .
thus , a suitable mass assignment window function should be localized both in real and fourier space . a good candidate that features these properties
is the scale function of the wavelet transformation .
the wavelet transformation has been previously introduced to astrophysical studies and has been applied successfully in the analysis of various astrophysical observations ( c.f . ,
fang & thews 1998 ) , e.g. , on the distributions of galaxies ( e.g. martinez et al . 1993 ; fang & feng 2000 ; yang et al .
, 2001 ; 2002a , b ; feng & fang 2004 ) , on the properties of ly@xmath102 absorption lines ( e.g. pando & fang 1997 ; meiksin 2000 ) , on the galaxy clusters ( e.g. , slezak et al .
1994 ; grebenev et al .
1995 ; gambera et al .
1997 ; schfer et al .
2005 ) , etc . here
, we introduce the scale function @xmath103 of the daubechies wavelet transformation for use in power spectrum measurements , which has the following properties ( e.g. , daubechies 1992 ) , @xmath104 @xmath105 and its fourier transform , @xmath106 , satisfies @xmath107 @xmath108 in this paper , we use the daubechies d12 and d20 scale functions ( daubechies 1988 , 1992 ) as our new mass assignment window functions , @xmath109 , which are shown in the left panel of fig . [
fig : assign_2 ] . in the right panel of fig .
[ fig : assign_2 ] , the squares of these two window functions in fourier space are shown as dotted and dashed lines , as indicated . for comparison , we also show in the right panel the ideal case of a top - hat fourier window function as the solid line .
the d12 and d20 window functions in fourier space @xmath82 resemble the ideal case very well , especially in the d20 case whose deviation from the ideal case at @xmath110 ( i.e. , @xmath111 ) is smaller than 2% .
note that these particular mass assignment window functions are different from the traditional schemes , e.g. ngp , cic and tsc in that : ( 1 ) they are not symmetric ; ( 2 ) they are not positive definite .
however these two features will not induce any undesirable consequences in our application .
first , since the overall shifting of the window function will not impact the amplitude of @xmath112 , the window function @xmath103 shown in the left panel of fig .
[ fig : assign_2 ] can be treated as symmetric components centered at @xmath113 and @xmath114 , respectively , with additional fluctuations at nearby grid cells .
second , the window function needs not necessarily be positive definite , as we are measuring the density contrast @xmath115 , and even the ideal window function @xmath116 is not positive definite .
before we turn to a discussion of their impact on measuring the true power spectrum , let us consider the computational cost for the mass assignment using the d12 and d20 scale functions . according to their real space behavior , at much better than 0.5% accuracy
, each mass particle should be distributed onto @xmath117 ( d12 ) or @xmath118 ( d20 ) grid cells , respectively , which is a factor of 10 or 20 times more than the tsc method with @xmath119 grid cells .
however , we argue that this cost is worthwhile given the following positive features of the daubechies assignment . first , according to eq .
( [ eq : d_k1 ] ) , the shot noise term in eq .
( [ eq : p_f ] ) for these mass assignment algorithms is , @xmath120 second , by comparing the fourier - space behaviors of the d12 and d20 functions with those of the traditional mass assignment methods , ngp , cic and tsc , it becomes clear that the three sampling effects of smearing , alias and anisotropy are greatly suppressed .
moreover , for the d20 window function , if we only measure the power spectrum up to @xmath3 , we do not need to take into account any of those three kinds of effects explicitly , because the true power spectrum is recovered with better than 2% accuracy !
another very important aspect is that such a mass assignment scheme can be fruitfully applied to the measurement of the higher - order spectra using an fft .
for instance , in measuring the bi - spectrum using an fft with the d20 mass assignment , we do not need to consider the sampling effects up to @xmath3 at all , since here the bi - spectrum can be measured directly with an accuracy level better than 3% .
note that so far there is no other approach known to accurately correct for the sampling effects in measuring the bi - spectrum with an fft .
we defer an application of our new technique and a theoretical modeling of the higher order spectra to a forthcoming paper .
having discussed the impact of the mass assignment window functions on the measurement of the true power spectrum using an fft , and having introduced the d12 and d20 scale functions , we proceed to demonstrate their performance when applied to the measurement of the mass power spectrum of a large n - body simulation . here we briefly describe the simulation , the millennium run , used for this project .
the millennium run is a very large dark matter simulation of the concordance @xmath121cdm cosmology with @xmath122 particles in a periodic box of @xmath123mpc on a side ( springel et al .
the simulation was carried out by the virgo consortium using a customized version of the gadget2 code .
the cosmological parameters used in this simulation are @xmath124 , @xmath125 , @xmath126 , @xmath127 , @xmath128 , and @xmath129 . for our test investigation
, we randomly select 10% of the dark matter particles ( because of practical limits in computer memory ) and measure their power spectra using the different window functions we described in the previous section .
to measure the power spectrum , we employ an fft of the density distribution of dark matter particles assigned to a grid with @xmath130 cells using the mass assignment algorithms discussed in section [ sec_window ] .
thus the corresponding nyquist wavenumber is @xmath131 . in the upper - left panel of fig [ fig : ps ]
, we show the fft power spectra measured using the traditional mass assignment functions , ngp , cic and tsc , where only the shot noise term has been subtracted . in this figure , the power spectrum is presented in terms of @xmath132 . for comparison , we show the theoretical prediction of the non - linear power spectrum by smith et al .
( 2003 ) as the solid line , based on the millennium run s cosmological parameters . obviously , because of the sampling effects we discussed in section [ sec : win1 ] , the power spectra are quite different at @xmath133 ( @xmath134 ) .
the power spectra measured without correcting the sampling effects , especially for the tsc method , significantly under - predict the true power spectrum . using the methods proposed by jing ( 2005 ) , we can iteratively correct for the sampling effects and extract estimates of the true power spectrum .
the corrected power spectra for the ngp , cic and tsc mass assignment methods are shown in the lower - left panel of fig . [
fig : ps ] . comparing these power spectra among themselves and with the ` halofit ' prediction of smith et al .
( 2003 ) , we are convinced that the true power spectrum is well recovered at all scales @xmath135 , and is roughly consistent with the prediction by smith et al .
now we turn to use the daubechies scale functions d12 and d20 as our mass assignment window functions .
the resulting power spectra after correcting for the shot noise term @xmath40 are shown in the upper - right panel of fig .
[ fig : ps ] , as indicated . without any correction for the sampling effects ,
the measured power spectra look very nice and match the theoretical predictions by smith et al .
( 2003 ) on all scales up to @xmath135 .
this is very different from the results shown in the upper - left panel of fig .
[ fig : ps ] for the classical assignment functions .
in fact , at a low resolution view , there is no visible difference between these results and the corrected measurements shown in the lower - left panel of fig .
[ fig : ps ] . finally , we take more accurate comparisons between the power spectra measured with these different methods by showing their ratios with respect to the ` halofit ' prediction of smith et al .
the de - convolved power spectra based on the ngp , cic and tsc mass assignments , the directly measured power spectra using d12 and d20 mass assignments are ploted together for comparison in the bottom right panel of fig . [
fig : ps ] . here
are a few observations that can be made : ( 1 ) the three de - convolved power spectra are very well consistent with each other at a level better than 2% at @xmath136 , and at a level of about 5% at @xmath137 ; ( 2 ) the directly measured power spectra based on the d12 and d20 ( the latter slightly better ) functions have an accuracy of better than 2% at @xmath136 and at a level of about 10% at @xmath138 ; ( 3 ) there is about 20% under - prediction on large scales ( with @xmath139 ) and 5% over - prediction on small scales by smith et al .
( 2003 ) for the power spectrum of the millennium simulation .
according to these findings , we may conclude that both the deconvolution method and the direct measurement based on the daubechies scale functions , especially for d20 , can recover the _ true _ power spectrum with better than 2% accuracy at @xmath140 .
moreover , as a conservative prediction , the bi - spectrum can be measured at a level better than 3% at @xmath141 if the d20 window function is used in the mass assignment for the fft .
this should be very useful for accurate studies of the bi - spectrum .
to quantify the large - scale structure in the distributions of a large population of dark matter particles or galaxies , one may measure their power ( or higher order ) spectra using a fft
. however , the required mass assignment onto the points of the fft - grid can introduce sampling effects in the measured power spectra .
most of these effects have been noticed and discussed in the literature before ( e.g. , baumgart & fry 1991 ; jing & valdarnini 1993 ; jing 2005 ) . among these , jing ( 2005 )
was the first to use an iterative correction method to compensate for all of these sampling effects , especially the alias effect . in this paper
, we follow jing ( 2005 ) and discuss the impact of the mass assignment on measuring the power spectrum with an fft .
there are two components that the employed window function can impact : one is the shot noise term and the other is the term involving the true power spectrum . with respect to the influence on the true power spectrum term , there are three different sampling effects that need to be considered : the smearing effect , the aliasing effect and the anisotropy effect . rather than trying to deconvolve for the sampling effects ,
we propose to use a special window function : the daubechies wavelet scale function that can minimize these sampling effects .
in particular , the d12 and d20 scale functions considered here are compact in real space , which allows a fast mass assignment onto the grid cells , while at the same time their top - hat like shape in fourier space leads only to very small sampling effects .
according to the fourier transform @xmath82 of the d20 function , at @xmath142 , all the sampling effects induced by the mass assignment can only affect the measured power spectrum at less than a level of 2% .
this is confirmed by the tests we carried out with the millennium run simulation .
more importantly , as a conservative prediction , the new method proposed here can measure the bi - spectrum of dark matter particles at better than 3% accuracy for @xmath136 , without the need to apply any correction for the sampling effects , apart from a simple substraction of the shot noise term .
we thank yipeng jing for helpful discussions , olaf wucknitz and the anonymous referee for helpful comments that greatly improved the presentation of this paper .
this work is supported by the _
one hundred talents _ project , shanghai pujiang program ( no .
07pj14102 ) , 973 program ( no .
2007cb815402 ) , 863 program ( no .
2006aa01a125 ) , the cas knowledge innovation program ( grant no .
kjcx2-yw - t05 ) and grants from nsfc ( nos .
10533030 , 10673023 , 10373012 , 10633049 ) . | in measuring the power spectrum of the distribution of large numbers of dark matter particles in simulations , or galaxies in observations , one has to use fast fourier transforms ( fft ) for calculational efficiency .
however , because of the required mass assignment onto grid points in this method , the measured power spectrum @xmath0 obtained with an fft is not the true power spectrum @xmath1 but instead one that is convolved with a window function @xmath2 in fourier space . in a recent paper , jing ( 2005 ) proposed an elegant algorithm to deconvolve the sampling effects of the window function and to extract the true power spectrum , and tests using n - body simulations show that this algorithm works very well for the three most commonly used mass assignment functions , i.e. , the nearest grid point ( ngp ) , the cloud in cell ( cic ) and the triangular shaped cloud ( tsc ) methods . in this paper , rather than trying to deconvolve the sampling effects of the window function , we propose to select a particular function in performing the mass assignment that can minimize these effects .
an ideal window function should fulfill the following criteria : ( i ) compact top - hat like support in fourier space to minimize the sampling effects ; ( ii ) compact support in real space to allow a fast and computationally feasible mass assignment onto grids .
we find that the scale functions of daubechies wavelet transformations are good candidates for such a purpose .
our tests using data from the millennium simulation show that the true power spectrum of dark matter can be accurately measured at a level better than 2% up to @xmath3 , without applying any deconvolution processes .
the new scheme is especially valuable for measurements of higher order statistics , e.g. the bi - spectrum , where it can render the mass assignment effects negligible up to comparatively high @xmath4 . |
of the more than 235 extrasolar giant planets ( egps " ) discovered in the last twelve years , 22 are transiting their primaries .
the most thoroughly studied transiting egp is hd 209458b ( henry et al . 2000 ; charbonneau et al .
2000 ; brown et al . 2001
; melo et al . 2006 ; santos , israelian , & mayor 2004 ; knutson et al .
2007a ) . for transiting egps , not only
do we measure the planet s radius , but the @xmath1 ambiguity of radial - velocity studies is resolved to yield the planet s mass .
moreover , with precision photometry , the wavelength - dependence of the transit radii can provide a measure of a planet s atmospheric composition ( burrows et al .
2000 ; hubbard et al . 2001
; fortney et al . 2003 ; barman 2007 ) . in this way
, sodium ( charbonneau et al . 2002 ) and water ( barman 2007 ) have been identified in hd 209458b and water has been identified in hd 189733b ( tinetti et al . 2007 ) and in tres-1 ( burrows , hubeny , & sudarsky 2005 ) . moreover , using the micro - satellite most , high - precision optical photometry has constrained ( perhaps , measured ) hd 209458b s geometric albedo ( rowe et al . 2006 , 2007 ) .
it is very low ( @xmath24.0@xmath3% ) , in keeping with the predictions of sudarsky , burrows , & pinto ( 2000 ) when the alkali metals , and not clouds , dominate absorption in the atmosphere and rayleigh scattering dominates scattering .
this makes hd 209458b darker in the optical than most of the objects of our solar system and also suggests that the associated contrast ratios are not optimal for characterizing egps . however , for hot , close - in egps such as hd 209458b , the planet - star contrast ratios in the mid - infrared are much more favorable than in the optical ( burrows , sudarsky , & hubeny 2003 ; burrows , sudarsky , & hubeny 2004 ; burrows 2005 ) , oftimes exceeding 10@xmath4 , and such contrasts are within reach of the infrared space telescope _ spitzer _
( werner & fanson 1995 ) . using its irac and mips cameras and the irs spectrometer , one can now measure the summed light of the planet and the star in and out of secondary eclipse ( when the star occults the planet ) and , from the difference , determine the planet s spectrum at superior conjunction .
this has led to a breakthrough in the study of exoplanets and a means to probe their chemistry and atmospheres .
hence , for the close - in egps in the near- to mid - infrared , and without the need to separately image planet and star , the direct detection of planetary atmospheres via low - resolution spectroscopy and precision ir photometry is now a reality ( charbonneau et al .
2005 ; deming et al .
2005,2006,2007 ; grillmair et al .
2007 ; richardson et al . 2007 ; harrington et al . 2006,2007 ; knutson et al . 2007b , c ; cowan et al .
2007 ) . to date , secondary eclipse fluxes in the irac and mips channels have been measured for five transiting egps ( hd 189733b , tres-1 , hd 209458b , hd 149026b , and gj 436b ) , but in this paper we focus on the interpretation of the hd 209458b data . a more comprehensive paper on our numerical technique and theoretical fits to other secondary eclipse data , as well as to phase light curve measurements , is in preparation ( burrows , budaj , & hubeny 2007b ) . for hd 209458b , not only
is there a complete set of measurements at secondary eclipse using irac and mips , but using the irs spectrometer , a low - resolution spectrum between @xmath27.5 @xmath0 m and @xmath215 @xmath0 m has now been obtained ( richardson et al .
2007 ; swain et al . 2007 ) .
excitingly , the new irac data of knutson et al .
( 2007c ) trace out a positive bump in the @xmath23.6@xmath0 m to 8 @xmath0 m spectral region , which we interpret via detailed modeling as a water _ emission _ feature .
a water absorption feature was anticipated ( burrows , hubeny , & sudarsky 2005 ; burrows , sudarsky , & hubeny 2006 ; fortney et al .
2005 ; barman , hauschildt , & allard 2005 ) .
therefore , this is strong evidence for a thermal inversion in the atmosphere of hd 209458b which has flipped water absorption features into emission features .
we now provide the comparison between theory and the observations and the resulting preliminary analysis .
three models with an upper - atmosphere absorber in the optical ( and with the resulting stratospheres ) for the planet / star flux ratios of hd 209458b at secondary eclipse are portrayed in fig .
the purple , green , and red models are for redistribution parameters ( p@xmath5 ; burrows , sudarsky , & hubeny 2006 ) of 0.3 , 0.4 , and 0.5 , respectively , but are otherwise the same . for each model , an extra absorber of uncertain provenance
is placed at altitude below pressures of 25 mbars .
the total monochromatic optical depth of this layer is @xmath23 .
a new redistribution algorithm that introduces a heat sink on the dayside between 0.01 and 0.1 bars , and a corresponding source on the nightside , is employed .
this algorithm is explained in burrows , budaj , & hubeny ( 2007b ) . following the suggestion of hubeny , burrows , & sudarsky ( 2003 ) in their atmosphere bifurcation study ,
we have also calculated a model suite ( not shown ) with equilibrium tio and vo in the stratosphere . with a total planck - mean optical depth of tio / vo below @xmath2 30 mbars of @xmath23.1 ,
these models are quite similar to those with inversions depicted in fig .
[ fig1 ] with the same p@xmath5s .
figure [ fig1 ] also provides a representative default model ( black ) without a stratospheric absorber , but with p@xmath5 = 0.3 , along with all the relevant hd 209458b measurements to date .
the default model represents the previous predictions for atmospheres with monotonically decreasing temperatures and no inversions , though the new redistribution algorithm alluded to above was employed .
the most salient observational constraints for our current purposes are the geometric albedo in the optical from most ( rowe et al .
2006,2007 ) , a k - band upper limit using irtf / spex from richardson , deming , & seager ( 2003 ) , a mips/24-@xmath0 m photometric point from deming et al .
( 2005 ) , a low - resolution _ spitzer_/irs spectrum from richardson et al .
( 2007 ) , and , most importantly , photometric points in irac channels 1 through 4 from knutson et al .
( 2007c ) .
richardson et al . ( 2007 ) suggest that there is evidence in the irs data for two spectral features : one near 7.78 @xmath0 m and one near 9.67 @xmath0 m .
however , we think the data are too noisy to draw this conclusion and await the next generation of observations to test it .
moreover , we note that richardson et al .
( 2007 ) normalize their data to the knutson et al .
( 2007c ) irac 4 point , but that due to the noisiness of the irs data near 8 @xmath0 m this normalization deserves a second look . the 1-@xmath6 optical albedo limit from rowe et al .
( 2007 ) is @xmath28.0% , a very low number . for comparison ,
the geometric albedo for jupiter is @xmath240% .
however , such a low number was predicted due to the prominence in the optical of broadband absorptions by the alkali metals sodium and potassium in the hot atmospheres of irradiated egps ( sudarsky et al .
2000 ; their class iv " ) .
alkali metals are not found in jupiter s atmosphere .
the associated planet - star flux ratios are @xmath210@xmath7@xmath810@xmath9 .
a low albedo due to absorption by the na d and k i resonance lines is consistent with the identification of sodium in the atmosphere of hd 209458b using hst / stis transit spectroscopy ( charbonneau et al .
2002 ) , though the magnitude of the sodium feature has yet to be fully explained .
both these datasets suggest that any clouds that might reside in the atmosphere of hd 209458b are thin or mostly absorbing .
a thick scattering cloud layer would reflect light rather efficiently and lead to a high albedo , which is not seen .
the extra absorber we introduce could in fact be a cloud , but in that case it must be comprised of particles with a low scattering albedo .
this would seem to eliminate forsterite and enstatite .
[ fig1 ] demonstrates , the low upper limit of richardson , deming , & seager ( 2003 ) in the k band that was problematic in the previous theory ( burrows , hubeny , & sudarsky 2005 ; fortney et al .
2005 ; barman , hauschildt , & allard 2005 ) is comfortably consistent with the models with an extra upper - atmosphere absorber in the optical , particularly for higher values of p@xmath5 .
moreover , the peak near the irac 1 channel ( @xmath23.6 @xmath0 m ) in the previous model without an inversion ( burrows et al . 2006 ) is reversed with the extra absorber into a deficit that fits the knutson et al .
( 2007c ) point .
importantly , the theory without an extra absorber at altitude predicts that the planet - star flux ratio in the irac 2 channel should be lower than the corresponding ratio at irac 1 .
however , with the extra absorber the relative strengths in these bands are reversed , just as are the knutson et al .
( 2007c ) points for hd 209458b .
this reversal is a clear signature of a thermal inversion in the low - pressure regions of the atmosphere .
figure [ fig2 ] depicts the corresponding temperature - pressure profiles and the thermal inversion at low pressures introduced by the presence of an extra absorber in the optical .
it also indicates the approximate positions of the effective photospheres for photons in the irac and mips channels . the fact that in the new models the irac 1 photosphere is cooler than the irac 2 photosphere is a key to the observed behavior of the knutson et al .
( 2007c ) data .
[ fig1 ] indicates that the models with a stratosphere fit the irac 1 , 2 and 4 flux points quite well .
however , the relative height of the irac 3 point near 5.8 @xmath0 m is difficult to fit , while maintaining the good fits at the other irac points and consistency with the richardson , deming , & seager ( 2003 ) limit . as fig .
[ fig2 ] shows , the positions of the irac 3 and irac 4 photospheres are generally close to one another .
this makes it difficult to have effective brightness " temperatures that are as different as are implied by the irac data .
nevertheless , irac 2 , 3 , and 4 together trace out a peak , whereas in the default theory an absorption trough was expected .
note in fig .
[ fig1 ] the significant difference in the planet - star flux ratio at these wavelengths between the old default model and the new models with a stratospheric absorber .
this is the spectral region of a strong ro - vibrational band of water and a comparison between our new hd 209458b models and the data indicates that this feature is in emission . the irs data of richardson et al .
( 2007 ) are normalized to the knutson et al .
( 2007c ) irac 4 point , so the good fit there is not independent .
nevertheless , the irs data have a flattish slope between @xmath27.5 @xmath0 m and @xmath214 @xmath0 m that is mildly inconsistent with the slight rise of our new models .
in addition , the 24-@xmath0 m mips point obtained by deming et al .
( 2005 ) is lower than these models .
however , the flux at this point is being reevaluated ( d. deming , private communication ) . if the new value at 24 @xmath0 m is , as suggested , @xmath20.0033@xmath10 , then our new model(s ) with inversions fit here as well ( see fig .
[ fig1 ] ) .
be that as it may , the good qualitative and quantitative fits for the k band limit , three of the four irac channels , and the most albedo limit ( not shown ) lend credence to our overall model and conclusions . concerning the dayside atmosphere of hd 209458b ,
there is an extra absorber in the optical at altitude , there is a pronounced thermal inversion ( see fig .
[ fig2 ] ) , and water is seen in emission .
we find that a consistent fit to all the hd 209458b data at secondary eclipse requires that the atmosphere of hd 209458b have a thermal inversion at altitude and a stratosphere .
this inversion is caused by the capture of incident optical stellar flux by an extra absorber of currently uncertain origin that resides at low pressures .
the irac data of knutson et al .
( 2007c ) can not be fit by the effects of dayside heat redistribution cooling alone .
a consequence of stratospheric heating and temperature inversion is the flipping of water absorption features into emission features from the near- to the mid - infrared ( hubeny , burrows , & sudarsky 2003 ; burrows , sudarsky , & hubeny 2006 ) , and this seems to explain all the current hd 209458b data .
hence , contrary to the interpretation of richardson et al .
( 2007 ) , the flatness or slight rise of their irs spectrum near @xmath27.8 @xmath0 m ( fig .
[ fig1 ] ) in fact supports the presence of abundant atmospheric water , because this region is at the edge of a strong water band in emission .
our inference of the presence of water in abundance in the atmosphere of hd 209458b is also consistent with the conclusion of barman ( 2007 ) using transmission spectroscopy .
if the extra absorber is in the gas phase , and there is no cloud , then our new models are easily consistent with the low albedo derived by rowe et al .
( 2006,2007 ) .
if the extra absorber is a cloud , the cloud particles must have a low scattering albedo and can not be very reflecting
. this would seem to rule out forsterite or enstatite clouds , but does not necessarily rule out iron clouds . as shown by hubeny , burrows , & sudarsky ( 2003 ) and burrows , sudarsky , & hubeny ( 2006 ) , strongly irradiated atmospheres can experience a solution bifurcation to an atmosphere with an inversion for which the water features are flipped from absorption to emission . in those papers ,
the absorber was gas - phase tio / vo , which in equilibrium can exist at low pressures at altitude and not just at high temperatures at depth .
if the extra absorber were tio / vo , there would be fewer free parameters , but it was thought that the cold - trap " effect would quickly deplete the upper atmosphere of tio / vo and ensure the default atmospheric solution without an inversion .
however , this has not been proven , particularly when atmospheric mass loss is ongoing , as we know to be the case for hd 209458b ( vidal - madjar et al .
2003,2004 ) .
the same arguments hold for iron clouds , although since absorption by iron particles is not restricted to the optical , this solution is sub - optimal .
one can also speculate that the severe irradiation regime of some close - in egps might create non - equilibrium compounds through photolysis , such as the tholins or polyacetylenes discussed in the more benign contexts of solar - system bodies , that could serve as the extra absorber we deduce exists in the atmosphere of hd 209458b .
clearly , what the high - altitude absorber actually is remains to be seen , but one can speculate that its presence is more likely in the atmospheres of the most strongly irradiated egps .
most strongly irradiated " actually means is not yet clear , but the flux at the substellar point on hd 209458b is @xmath2@xmath11 erg @xmath12 s@xmath13 . the corresponding numbers for ogle - tr-10b , ogle - tr-56b , ogle - tr-132b , tres-2 , tres-3 , wasp-1b , xo-3b , hat - p-1b , and hd 149026b are higher ( burrows et al .
the corresponding number for tres-1 is lower ( @xmath20.43@xmath14 erg @xmath12 s@xmath13 ) and this planet shows good evidence for water in _ absorption _ ( burrows , hubeny , & sudarsky 2005 ) .
in addition , the irs spectrum of hd 189733b of grillmair et al .
( 2007 ) seems consistent with a more canonical water absorption feature shortward of @xmath28 @xmath0 m .
its substellar flux is @xmath20.47@xmath14 erg @xmath12 s@xmath13 .
so , if stellar flux at the planet is an indicator , we may have a handle on which planets reside in the transition region between manifesting water absorption or emission features ( and inversions ) , and where the relative irac 1/irac 2 strengths start to flip . given their substellar fluxes , hd 189733b , xo-1b , xo-2b , and/or wasp-2b may be links . though we find a weak dependence on metallicity , non - equilibrium chemistry and cloud formation may have stronger dependences
. therefore , there are still numerous parameters to address . as fig .
[ fig1 ] suggests , the planet / star ratios shortward of @xmath24 @xmath0 m are more sensitive to the heat redistribution parameter , p@xmath5 , than the corresponding ratios longward of @xmath28 @xmath0 m .
this suggests a shorter - wavelength strategy for designing close - in egp climate diagnostics .
in addition , models with inversions significantly boost the mid - infrared fluxes longward of @xmath215 @xmath0 m .
this boost is not the only signature of models with dayside stratospheres .
since the nightside flux at the same p@xmath5 is unlikely to be much altered by the fact of dayside inversion , if the dayside has such an inversion due to enhanced optical absorption at altitude , the day - night contrasts measured during an orbital traverse by a close - in egp will be larger than expected for a given p@xmath5 .
this possibility is relevant when interpreting the 24-@xmath0 m light curve of @xmath15 and b ( harrington et al .
perhaps , the large day - night contrast seen in this case does not require a small p@xmath5 .
an upper - atmosphere absorber in the optical may simultaneously explain the weaker - than - expected na d feature seen in transit by charbonneau et al .
( 2002 ) ( fortney et al . 2003 ) and the fact that the transit radius measured by richardson et al .
( 2006 ) at 24 @xmath0 m ( [email protected] r@xmath17 ) is smaller than the corresponding radius in the optical ( [email protected] r@xmath17 ; knutson et al .
in addition , it may help explain why hd 209458b s optical transit radius is as large as it is .
finally , the extra absorber may be related to the enhanced opacities for close - in egps discussed in the context of the egp radius models of burrows et al .
( 2007a ) .
whatever their actual roles , anomalous optical absorption at altitude and thermal inversions are now insinuating themselves as exciting new components of egp theory .
we thank drake deming , bill hubbard , maki hattori , mike cushing , and drew milsom for helpful discussions .
this study was supported in part by nasa grants nng04gl22 g and nnx07ag80 g and through the nasa astrobiology institute under cooperative agreement no .
can-02-oss-02 issued through the office of space science .
burrows , a. , sudarsky , d. , & hubeny , i. 2003 , published in the proceedings of the 14th annual astrophysics conference in maryland the search for other worlds , " eds . s. holt and d. deming , ( aip conference proceedings ) , held in college park , md , october 13 - 14 , 2003 , p. 143 .
burrows , a. , hubeny , i. , & sudarsky , d. , 2005 , 625 , l135 burrows , a. , sudarsky , d. & hubeny , i. 2006 , , 650 , 1140 ( astro - ph/0607014 ) burrows , a. , hubeny , i. , budaj , j. , & hubbard , w.b .
2007a , , 661 , 502 burrows , a. , hubeny , i. , & budaj , j. 2007b , in preparation charbonneau , d. , brown , t. m. , noyes , r. w. , & gilliland , r. l. 2002 , , 568 , 377 charbonneau , d. 2005 , , 626 , 523 deming , d. , seager , s. , richardson , l.j .
, & harrington , j. , 2005 , nature , 434 , 740 deming , d. , harrington , j. , seager , s. , richardson , l.r .
2006 , , 644 , 560 deming , d. , harrington , j. , laughlin , g. , seager , s. , navarro , s.b . ,
bowman , w.c . ,
& horning , k. 2007 , submitted to letters ( arxiv:0707.2778 ) fortney , j.j . , sudarsky , d. , hubeny , i. , cooper , c.s .
, hubbard , w.b . , burrows , a. , & lunine , j.i .
2003 , , 589 , 615 fortney , j.j . , marley , m.s . , lodders , k. , saumon , d. , & freedman , r.s .
2005 , , 627 , l69 grillmair , c.j . , charbonneau , d. , burrows , a. , armus , l. , stauffer , j. , meadows , v. , van cleve , j. , & levine , d. 2007 , , 658 , l115 harrington , j. , hansen , b. , luszcz , s. , seager , s. , deming , d. , menou , k. , cho , j. , & richardson , l. 2006 , science , 314 , 623 harrington , j. , luszcz , s. , seager , s. , deming , d. , & richardson , l.j .
2007 , nature , 447 , 691 knutson , h.a . ,
charbonneau , d. , noyes , r.w . ,
brown , t.m . , & gilliland , r.l .
2007a , , 655 , 564 knutson , h.a . ,
charbonneau , d. , allen , l.e . ,
fortney , j.j . ,
agol , e. , cowan , n.b . , showman , a.p . ,
cooper , c.s . , & megeath , s.t .
2007b , nature , 447 , 183 knutson , h.a . , charbonneau , d. , allen , l.e . ,
torres , g. , burrows , a. , & megeath , s.t .
2007c , submitted to melo , c. , et al .
2006 , , 460 , 251 richardson , l.j . , deming , d. , & seager , s. 2003 , , 597 , 581 richardson , l.j . , harrington , j. , seager , s. , & deming , d. 2006 , , 649 , 1043 richardson , l.j . , deming , d. , horning , k. , seager , s. , & harrington , j. 2007 , nature , 445 , 892 rowe , j.f . , matthews , j.m .
, seager , s. , kuschnig , r. , guenther , d.b . ,
moffat , a.f.j . ,
rucinski , s.m . ,
sasselov , d. , walker , g.a.h . , &
weiss , w.w .
2006 , , 645 , 1241 ( astro - ph/0603410 ) rowe , j.f . , et al .
2007 , submitted to santos , n.c . , israelian , g. , & mayor , m. 2004 , , 415 , 1153 swain , m. , bouwman , j. , akeson , r. , lawler , s. & beichman , c. 2007 , submitted to letters ( arxiv : astro - ph/0702593 ) tinetti , g. et al .
2007 , nature , 448 , 169 vidal - madjar , a. , lecavelier des etangs , a. , dsert , j .- m . , ballester , g.e . ,
ferlet , r. , hbrand , g. , & mayor , m. 2003 , nature , 422 , 143 | we find that a theoretical fit to all the hd 209458b data at secondary eclipse requires that the dayside atmosphere of hd 209458b have a thermal inversion and a stratosphere .
this inversion is caused by the capture of optical stellar flux by an absorber of uncertain origin that resides at altitude .
one consequence of stratospheric heating and temperature inversion is the flipping of water absorption features into _ emission _ features from the near- to the mid - infrared and we see evidence of such a water emission feature in the recent hd 209458b irac data of knutson et al .
in addition , an upper - atmosphere optical absorber may help explain both the weaker - than - expected na d feature seen in transit and the fact that the transit radius at 24 @xmath0 m is smaller than the corresponding radius in the optical .
moreover , it may be a factor in why hd 209458b s optical transit radius is as large as it is .
we speculate on the nature of this absorber and the planets whose atmospheres may , or may not , be affected by its presence . |
various optical phenomena associated with the optical pumping process , which have been comprehensively studied since the 60 s and described in many aspects in a famous review @xcite by w.happer , is revived nowadays in a new form within the field of quantum information and quantum computing .
the paramagnetic ground states of macroscopic atomic spin subsystems are considered now convenient physical objects for mapping and storing the quantum information in the quantum states of their collective angular momenta .
the faraday - type interference scheme was proposed for spin squeezing @xcite and for quantum communication between atomic ensembles @xcite .
the proposed ideas were realized in spin squeezing @xcite and in the entanglement @xcite experiments , where a quantum measurement on the light forward - scattered from atomic ensembles was used .
the same kind of off - resonant forward scattering combined with a quantum feedback was used in the recent demonstration of quantum memory for light @xcite .
applications to various quantum information protocols including a cat state generation @xcite and quantum cloning of light onto atoms @xcite have been proposed .
efficient generation of entanglement via multi - pass interaction have been also proposed @xcite .
theoretical modelling in the above mentioned papers was concerned with collective canonical variables for atoms which can be conveniently introduced for spin @xmath0 atoms .
however , the actual experiments were conducted using states with higher angular momenta . hence a theoretical model describing off - resonant interaction of light with realistic atoms with the angular momentum higher
then @xmath0 is required .
it is worth noting that the effect of a random faraday rotation due to atomic fluctuations was discussed and observed first in refs.@xcite more than twenty years ago in the context of demonstrating the advantages of light - beating method in atomic spectroscopy .
the importance of atomic polarization was later discussed in refs.@xcite where it was shown that the collective spin polarization of an atomic ensemble could essentially modify the quantum statistics of the outgoing off - resonant probe radiation .
quantum statistics of atomic spin variables was first experimentally observed via off - resonant forward scattering in @xcite . in the present paper
we develop the quantum theory of coherent forward scattering on an ensemble of polarized atoms with an arbitrary angular momentum .
we discuss the physical conditions under which the forward scattering can be properly described in terms of the effective hamiltonian , generalizing the semiclassical approach @xcite , so that both the field and atomic subsystems are treated as quantum objects .
based on the effective hamiltonian we derive the heisenberg - type equations of motion written for the spatial distribution of the stokes variables of the field and of the collective angular momentum of atoms .
the solution shows how the entanglement of the quantum fluctuations of light and atomic subsystems is formed .
we show that the entangling process can be understood in terms of a polariton - type spin wave induced in the atomic sample .
we support our discussion by numerical simulations , assuming the conditions close to those of the recent experiments @xcite .
the paper is organized as follows . in section [ s2 ]
we review the process of coherent forward scattering to show how the field heisenberg operators are transformed in interaction with multi - atom ensemble consisting of atoms with an arbitrary zeeman structure and for an off - resonant excitation on a dipole allowed optical transition . in section [ s3 ]
we extend the heisenberg formalism to the atomic subsystem and introduce the effective hamiltonian responsible for the coupled dynamics of an infinite number of variables associated with the local polarization of the weak probe light and with individual atomic spins .
the effective hamiltonian is further transformed in section [ s4 ] into another form by introducing the more convenient for the polarization sensitive processes irreducible tensor formalism . in section [ s5 ]
we perform mesoscopic averaging to arrive at the wave - type equations for the collective quantum variables of the light and atomic subsystems .
these equations are solved and discussed in section [ s6 ] in the context of the quantum entangling process .
in particular they are used to test the applicability of the spin one - half approximation for modelling the quantum non - demolishing measurement in an ensemble of alkali atoms with a higher angular momentum .
consider a system of @xmath1 identical atoms located in a finite volume and scattering coherent light of frequency @xmath2 .
the wave of light incident on the atomic ensemble is assumed to be weak enough that possible saturation effects in its interaction with the atoms are negligible .
the atomic ensemble , in general , represents an optically long ( thick in refraction , but thin in absorption ) medium for multiple off resonance scattering , but , on the average , the atoms are separated by distances much larger than the wavelength @xmath3 .
thus each atom is located , on average , in the radiation zone of its neighbors .
the interaction of the atoms with incident and multiply - scattered light is assumed to be of the dipole type and a proper description of multiple scattering has to be restricted to the rotating wave approximation . to follow as clearly as possible the analogy between classical and quantum descriptions of the coherent forward scattering it seems convenient to solve the quantum problem in the heisenberg picture .
in such an approach the basic process is the transformation of the electric field operators in a single scattering event .
this process is reviewed in appendix [ s2.1 ] , as an example of the transformation of the operator of the free electric field of a moving atom . as shown there , the basic result can be written in terms of the heisenberg - type microscopic maxwell equation , where polarization response - operator is given by a single point - like scatterer .
this allows us to make a subsequent generalization to the situation of an arbitrary number of atoms scattering light coherently in the forward direction .
we begin our discussion from the single particle microscopic problem , described by equation ( [ 2.10])(see the appendix [ s2.1 ] for details ) . for a system of @xmath1
identical scatterers randomly located in space it would be not so easy to generalize ( [ 2.10 ] ) to a multi - particle form allowing to follow precisely the heisenberg dynamics . for simplicity
we restrict the discussion to the case of the plain incident wave , as in the experiments with gas cells @xcite . for a transparent medium consisting of @xmath1 scatterers light
is scattered in any direction other than the forward direction only incoherently .
this means that for optically thin and transparent medium the probability to observe a photon randomly scattered in any solid angle different from the direction of light propagation is just a sum of partial probabilities of independent scattering by each atom . in the heisenberg formalism
this means that the non - forward propagating operator waves created in the medium by single or multiple scattering events are superposed with random phases to a roughly zero sum .
the outgoing flux associated with incoherent scattering is proportional to the total number of atoms , but stays small because of a negligible value of the off - resonant cross section . at the same time for light propagating in the forward direction in transparent medium
there is a very strong coherent enhancement of the scattering process .
since the doppler shift in the forward scattered modes disappears , see ( [ 2.8 ] ) , and the raman shift , caused by zeeman splitting , does not noticeably change the phase of these modes , the transmitted light reveals strong coherent superposition of all partial contributions associated with single and multiple scattering . for discussion of the off - resonant light - atom coherent scattering beyond the plain wave approximation see @xcite .
let us subdivide the joint dynamics of light and atomic subsystems into time increments , during which the number of accumulated incoherent scattering events is much less than the total number of atoms .
this simplifies our analysis and allows to consider only the modes coherently scattered in the forward direction . in the examples we are going to discuss below this type of scattering
is additionally stimulated by the propagating mode , which has the quasi - classical nature .
within the rotating wave approximation we introduce a carrier frequency @xmath4 , coinciding with the average frequency of propagating light , and define an averaged wavenumber @xmath5 associated with the full set of the modes propagating in the @xmath6-direction .
then the spectral bandwidth @xmath7 of the continuum of the field modes , contributing to the quantum operator expansion , should be chosen to be less than @xmath4 .
the actual spectrum of the incident radiation , centered at @xmath4 , has a narrower bandwidth @xmath8 , which is much less than the detuning from the resonance @xmath9 , where @xmath10 is the resonance frequency of the non - perturbed optical transition .
we will also assume that @xmath11 is much greater than natural linewidth @xmath12 , doppler shift @xmath13 , and raman shift @xmath14 .
we neglect the imaginary shift associated with the natural linewidth in the energy denominators . the dissipation process associated with the spontaneous decay
is neglected in our discussion ; the assumptions justifying this are discussed in section [ s6.c ] . for an analysis of the role of spontaneous emission for off - resonant entangling interaction
see @xcite . under the above mentioned assumpltions the basic expression for the polarization of any @xmath15-th atom contributing to the forward scattering ( [ 2.15 ] )
can be rewritten as follows @xmath16 where @xmath17is the polarizability tensor and @xmath18 is the crossection area of the light beam propagating through the medium in @xmath6-direction .
other notations are specified in appendix [ s2.1 ] . expression ( [ 2.17 ] ) gives the instantaneous value of the polarizability tensor for @xmath15-th atom , which depends on its exact heisenberg evolution to the moment @xmath19 .
the heisenberg operator of the electric field on the right hand side of equation ( [ 2.16 ] ) is assumed to be dressed in the perturbations by other atoms located in front of any selected @xmath15-th atom but non - perturbed by the selected atom itself .
here we use cartesian coordinates in the tensor notation and sum over each repeated index .
the indices @xmath20 take values @xmath21 or @xmath22 .
let us consider a thin mesoscopic layer located between @xmath6 and @xmath23 planes and containing a large number of atoms .
such an atomic sub - ensemble will scatter incoming field coherently in the forward direction by the collective polarization @xmath24where @xmath6 coordinate is confined inside the layer @xmath25 . if the medium is split in a number of thin layers along the propagation direction the light beam would be subsequently scattered by this layers in the forward direction .
then the `` coarse - grained '' dynamics of the field operators can be described by the following heisenberg - type macroscopic maxwell equation @xmath26where the polarization operator @xmath27 is subsequently given by expressions ( [ 2.16 ] ) - ( [ 2.18 ] ) .
we conclude this section by the following remark . the derived macroscopic maxwell equation is coupled to the corresponding heisenberg equations governing the dynamics of the atomic subsystem ,
see below , and it can not be extended up to an arbitrary time @xmath19 . its validity is restricted by ignoring the dissipation process of incoherent scattering .
that is why the averaging of the operator polarizability ( [ 2.17 ] ) gives only the refraction part of the real polarizability tensor of normal maxwell equations .
we will further discuss the self - consistency of the dynamical approach in section [ s6.c ] .
in this section we discuss how the atomic variables are modified via the interaction with the forward propagating light .
we describe the dynamics of slowly varying ground state operators , considering at first a single atom illuminated with an arbitrary quantized optical field .
then we generalize the problem to a macroscopic ensemble and introduce atomic collective variables .
consider an atom at the origin of the coordinate frame at the initial moment of time and drifting in space in such a way that during the scattering event it moves much less than a wavelength .
let us define an arbitrary dyadic - type operator for the ground state of this atom @xmath28where by the second equality we specify the atomic state more precisely and introduce the atomic quantum numbers : @xmath29 and @xmath30 are the ground state angular momentum and its projections on the direction of an external magnetic field oriented along the @xmath31 axis , which in general is different from the direction of the light propagation .
based on a perturbation theory we can expand the corresponding heisenberg operator up to the second order as follows @xmath32where @xmath33is the operator in the interaction representation and the second term @xmath34 \right]% \label{3.4}%\end{aligned}\]]is the second order correction induced by a dipole - type optical interaction , see ( [ 2.2 ] ) .
the integral ( [ 3.4 ] ) should be evaluated in the rotating wave approximation by keeping only the leading terms in the limit latexmath:[$t\gg products of the creation and annihilation field operators . in general such products
expand over all the spatial modes , but in reality only those modes , which will not vanish after the averaging over the initial state , are important .
these are planar modes propagating along the @xmath6-axis .
taking into account only the propagating modes , the expansion ( [ 3.2 ] ) can be rewritten as follows @xmath36\,+\,\ldots% \label{3.5}%\]]where we introduce an effective interaction hamiltonian in the interaction representation @xmath37 strictly speaking the electric field operators and the operator of the atomic polarizability tensor should be understood here as defined in the interaction representation and marked by the index zero .
for the electric field these operators @xmath38 are given by the expansion ( [ 2.1 ] ) , defined at @xmath39 , where we keep only the forward propagating modes .
the operator of the polarizability tensor @xmath40 is given by eq.([2.17 ] ) , transformed to the interaction representation , with @xmath4 and @xmath5 being the carrier frequency and wavenumber of the modes interacting with an atom .
as in the eq.([2.16 ] ) the tensor indices inside the definition of the effective hamiltonian ( [ 3.6 ] ) can run only two projections either @xmath21 or @xmath22 .
note that in our derivation of the effective hamiltonian we used a rather short time increment @xmath19 consistent with the perturbation theory approach ; with the assumption that the atom does not noticeably change its location during the scattering event . then there would be no difference between the interaction and the heisenberg representations in the evaluation of the integral ( [ 3.4 ] ) and in introducing the effective hamiltonian ( [ 3.6 ] ) .
based on a general principle of dynamical evolution we can straightforwardly generalize eq.([3.6 ] ) up to an arbitrary moment in time if we substitute there all the operators in the heisenberg representation .
we should also take into account a classical drift of the atom in space and consider @xmath41 coordinate as its actual location at moment @xmath19 .
consider now an ensemble consisting of many atoms scattering incident light coherently in the forward direction .
although the most interesting polariton - type solutions obtained further in the paper are destroyed by atomic motion , we include the motion in our model for the sake of generality
. however we ignore any possible correlations between their spatial motion and the internal state evolution .
then each atom is in the environment of the field scattered by the atoms located in front of it and coupled with such a dressed field via the partial effective hamiltonian ( [ 3.6 ] ) .
the full effective hamiltonian for the whole ensemble interacting with the propagating field in the heisenberg representation is given by @xmath42we preserve here the notation for the full effective hamiltonian , which was used in the previous equation in the case of one atom .
then operator ( [ 3.1 ] ) considered for each atom of the ensemble satisfies the following heisenberg equation @xmath43% \label{3.8}%\ ] ] where @xmath44 .
if we take into account the commutation relation between heiesenberg operators of the electric field propagating in the forward direction , which are given by @xmath45\;=\;% -\frac{2\pi\hbar\bar{\omega}}{s_0}\,\delta_{ki}\,\delta(z'-z)% \label{3.9}%\]]then the maxwell equation ( [ 2.19 ] ) can be also rewritten in terms of the effective hamiltonian @xmath46% \label{3.10}%\ ] ] this operator equation shows that , in turn , the state of the field existing at each spatial point of the medium is defined by the atomic operators in the heisenberg representation , according to their spatial distribution in space at a given time @xmath19 .
let us make the following remark concerning the definition of @xmath47-functions in ( [ 3.9 ] ) and in ( [ 2.16 ] ) and the validity of the dynamical equations in the form ( [ 3.10 ] ) .
our description of the quasi - resonant radiation propagating in disordered medium in the forward direction is based on the rotating wave approximation . according to this approximation the spectral bandwidth of the field @xmath7
is assumed to be at least less than carrier frequency @xmath4 .
such a truncation of the infinite field continuum makes possible to consider the commutation relations of truly heisenberg operators in the form ( [ 3.9 ] ) .
thus the @xmath47-functions here and in ( [ 2.16 ] ) should be correctly understood as distributed in a small mesoscopic area of the order @xmath48 .
this spatial scale is obviously longer than @xmath3 but still much shorter than the sample size or than any internal macroscopic scale associated with macroscopic susceptibilities of the medium .
it is also important to think about equations ( [ 3.10 ] ) as on approximation for more general multi - mode heisenberg - langevin type equation , where the damping processes associated with incoherent scattering would be taken into consideration . as was mentioned before , the validity of the output maxwell equations ( [ 2.19 ] ) , ( [ 3.10 ] ) is actually restricted by the narrow spectral domain @xmath8 associated with the spectrum of the probe light .
but in the process of transposing the field operators by means of the commutation rule ( [ 3.9 ] ) the whole field spectrum @xmath7 should be taken into account .
the coupled equations ( [ 3.8 ] ) and ( [ 3.10 ] ) considered together is the main result of this section .
they reveal the joint dynamics of the field and atomic subsystems interacting in the limit of non - saturating off - resonant optical excitation . being an example of heisenberg equations they are valid for arbitrary initial quantum state of the field and of the atomic ensemble .
the main restriction comes from the model of lossless coherent scattering .
but even with such a simplification these equations are quite complicated since they are operator equations for an _ infinite number _ of the field and atomic variables . in the following sections we identify and discuss special conditions when equations ( [ 3.8 ] ) and ( [ 3.10 ] ) could be converted to a
_ finite number _ of heisenberg - type equations for collective variables of the field and atomic subsystems .
the set of dyadic - type operators ( [ 3.1 ] ) for each atom of the ensemble can be replaced with another set of operators @xmath49being the linear combination of original operators weighted with clebsh - gordan coefficients @xmath50 , the projectors @xmath51 become irreducible tensor operators , which possess the simplest properties with respect to rotational transformations , see @xcite .
the linear transformations ( [ 4.1 ] ) are consistent with any coordinate frame .
let us abandon the choice of the frame with @xmath31-axis along magnetic field , used in the basic definition ( [ 3.1 ] ) , and return to the original frame with the @xmath6-axis along the propagating beam , which is more natural for the further discussion of the effective hamiltonian approach . for large frequency
detuning the zeeman splitting in the energy denominators can be neglected compared with the average detuning @xmath52 , where @xmath53 is the transition frequency between the ground and excited states characterized by the angular momenta @xmath29 and @xmath54 respectively .
then the effective hamiltonian can be expressed as the sum of three terms @xmath55where dependence on time emphasizes that the contributing atom and field operators are taken in the heisenberg representation . the first term in ( [ 4.2 ] ) couples the atomic population of the whole zeeman multiplet and their longitudinal alignment with the full photon flux of propagating light @xmath56\;%
\hat{\xi}_0\left(z_a(t),t\right)% \label{4.3}%\end{aligned}\]]where @xmath57is the stokes operator of the total photon flux at the spatial point @xmath6 .
the isotropic polarizability of an atom is given by @xmath58where @xmath59 is the reduced matrix element of the atomic dipole moment for @xmath60 transition .
the alignment component of the atomic polarizability is given by eq.([4.13 ] ) below .
the second term in eq.([4.2 ] ) couples the gyrotropic or orientation component of the atomic ensemble with the stokes component responsible for circular polarization of propagating light @xmath61where
@xmath62% \label{4.7}%\end{aligned}\]]is the stokes operator associated with circular polarization .
it is defined in terms of the photon flux operators at a spatial point @xmath6 and shows imbalance between the right - hand ( @xmath63 ) and the left - hand ( @xmath64 ) polarizations of the field .
the orientational polarizability of an atom is given by @xmath65as follows from the maxwell equation ( [ 3.10 ] ) in the classical limit , the term @xmath66 completely defines the faraday rotation or other gyrotropy effects existing in a bulk medium .
the third term in eq.([4.2 ] ) couples the alignment components of the atomic ensemble with the remaining two linear polarized type stokes components of propagating light @xmath67% \label{4.9}%\end{aligned}\ ] ] where @xmath68% \nonumber\\% \hat{t}_{\xi\eta}^{(a)}(t)&=&% -\frac{1}{2i}\left[\hat{t}_{2 - 2}^{(a)}(t)\,-\,% \hat{t}_{22}^{(a)}(t)\right]% \label{4.10}%\end{aligned}\]]the stokes operator @xmath69 , showing imbalance between the photon fluxes of the modes linearly polarized along @xmath21 and @xmath22 axes , is given by @xmath70% \label{4.11}%\end{aligned}\]]the stokes operator @xmath71 , showing imbalance between photon fluxes of the modes linearly polarized along @xmath72 and @xmath73 axis rotated with respect to @xmath21 and @xmath22 directions by @xmath74 angle , is given by @xmath75% \label{4.12}%\end{aligned}\ ] ] the alignment component of the atomic polarizability is defined as follows @xmath76as follows from the maxwell equation ( [ 3.10 ] ) considered in its classical limit the term @xmath77 is responsible for the optical birefringence effects with respect to either @xmath21 and @xmath22 or @xmath72 and @xmath73 directions
. equation ( [ 3.8 ] ) in the irreducible representation transforms into @xmath78% \nonumber\\% & & + \;\frac{i}{\hbar}\left[\hat{\cal{h}}_{eff}(t),\,% \hat{t}_{kq}^{(a)}(t)\right]% \label{4.14}%\end{aligned}\ ] ] where @xmath79 is the unperturbed hamiltonian responsible for the interaction with the external magnetic field .
recall here that in general the direction of the magnetic field does not coincide with the @xmath6-axis .
the set of equations ( [ 4.14 ] ) for @xmath80 is equivalent to the set of equations ( [ 3.8 ] ) but it is written for more convenient physical observables .
the irreducible components of a low rank , which contribute to the effective hamiltonian and couple to the stokes observables of the propagating light , allow for a more transparent interpretation than the original projector operators ( [ 3.1 ] ) .
the zero rank component @xmath81 is the heisenberg operator of the total population of an a - th atom in its ground state .
it can be straightforwardly verified that the right hand size of ( [ 4.14 ] ) is equal to zero in this case and the zero rank irreducible operator is , in fact , the identity operator .
the first rank component @xmath82 , being the heisenberg operator of atomic orientation , is equivalent to the vector of the atomic angular momentum .
this vector undergoes dynamical evolution ( regular precession and coupling with the field variables ) caused by the perturbation of the atomic ground state by the propagating light .
the second rank component @xmath83 , being the heisenberg operator of atomic alignment , is equivalent to the ground state quadrupole moment of the atom .
the quadropole tensor also undergoes the dynamical evolution caused by interaction with the propagating light .
other higher rank irreducible components are important only as long as their evolution is dynamically coupled with the evolution of the lower rank components in the complete set of eqs.([4.14 ] ) .
the equations ( [ 4.14 ] ) have to be solved together with the maxwell equations .
let us introduce the slowly varying amplitudes of the heisenberg field operators @xmath84 by substituting these expressions into equation ( 3.10 ) and into its hermitian conjugated form , we obtain the following first order differential equations for slowly varying amplitudes @xmath85% \hat{\epsilon}_i(z , t)&=&% \frac{i}{\hbar c}\left[\hat{\cal{h}}_{eff}(t),\,% \hat{\epsilon}_i(z , t)\right]% \nonumber\\% \left[\frac{\partial}{\partial z}+\frac{1}{c}% \frac{\partial}{\partial t}\right]% \hat{\epsilon}^\dagger_i(z , t)&=&% \frac{i}{\hbar c}\left[\hat{\cal{h}}_{eff}(t),\,% \hat{\epsilon}^\dagger_i(z , t)\right]% \label{4.16}%\end{aligned}\]]in such heisenberg - transport equations , as well as in the dynamical equations ( [ 4.14 ] ) , we consider time @xmath19 in a coarse - grain temporal scale much longer than @xmath86 and the coordinate @xmath6 on a coarse - grain spatial scale much longer than @xmath3 .
any changes in the atomic subsystem and displacement of atoms during the time increments comparable with @xmath86 as well as any changes of the slow varying field operators on the scale of a few @xmath3 are ignored .
we do not actually know the exact behaviour neither for the atom nor for the field heisenberg operators , but we can approximately display their averaged behaviour by solution of coupled equations ( [ 4.14 ] ) and ( [ 4.16 ] ) . instead of the heisenberg - type equations for slowly varying amplitudes @xmath87 and @xmath88 similar equations written for the stokes operators can be introduced @xmath89% \hat{\xi}_i(z , t)\;=\;%
\frac{i}{\hbar c}\left[\hat{\cal{h}}_{eff}(t),\,% \hat{\xi}_i(z , t)\right]% \label{4.17}%\]]where @xmath90 . all the terms appearing on the right hand sides of equations ( [ 4.14 ] ) and ( [ 4.17 ] ) can be expressed via the stokes operators .
moreover it can be straightforwardly verified that @xmath91 stays unchanged as a function of @xmath92 , because of the conservation of the number of photons in the forward scattering process . in turn
this means that the first term in eq.([4.3 ] ) , i. e. the isotropic component of the effective hamiltonian , can be omitted in ( [ 4.17 ] ) since it commutes with all the stokes operators . as follows from the derived equations the main obstacle on the way to convert the infinite number of operator equations to the system of truncated equations , written for the collective atomic variables , coupled with the integral or averaged field variables , comes from the spatial dependence of the interaction process .
such a dependence is caused by anisotropic terms in the effective hamiltonian , which lead to spatially varying entanglement of different polarization modes of light with atomic spins along the propagation path .
the spatial profile of the heisenberg - type stokes operators could be considered as uniform in @xmath93 component and as accumulating the collective fluctuations of atomic spins in complementary @xmath94 component only in one special case of a pure faraday effect . in this case
the anisotropic components of the local susceptibilities of the medium would be zero on average and would exist as fluctuations acting only on the @xmath94 stokes component via random faraday rotation .
this can be true if @xmath95 , @xmath96 and @xmath97 operators would be zero on average along the whole interaction cycle .
but even if we neglect the repopulation optical pumping mechanism , coming from incoherent scattering , it would be not so easy to show any realistic example of a proper atomic transition satisfying this condition . such an example could be @xmath98 or @xmath99 atomic transitions .
then there is no quadropole moment in the ground state in principle and the ensemble consists of the atoms perfectly polarized ( oriented ) in the direction orthogonal to the propagation direction of the probe light .
the atom with spin one - half in its ground state does exist in reality ( @xmath100he for example ) , but it is not a convenient object for the polarization sensitive experiments . in the earlier work on entanglement and quantum information protocols with atomic continuous variables
@xcite realistic atoms were modelled as spin one half systems .
one of the goals of this paper is to analyze the applicability of such approximation . below we derive equations of motion in their general form and discuss their possible conversion to the finite number of wave - type equations written in terms of spatially distributed collective variables .
as the most important practical example , we will further discuss the light propagation through atomic ensemble prepared originally in the coherent spin state , see figure [ fig.1 ] .
we assume that atoms fill the cylindrical volume with the cross section @xmath18 coinciding with the cross section of the probe light beam .
the atomic ensemble is located in the homogeneous magnetic field with the direction orthogonal to the direction of the propagating light .
originally atoms are perfectly oriented along the magnetic field in the @xmath21 direction .
probe light is in a coherent state linearly polarized along the @xmath21 direction .
such a geometry is common , e.g .
, for experiments aiming at the quantum state teleportation between field and atomic spin subsystems @xcite .
let us consider first the evolution of the stokes components .
based on the commutation relation @xmath101\;=\;% 2i\varepsilon_{ijk}\,c\,\delta(z - z')\hat{\xi}_k(z , t)% \label{5.1}%\ ] ] where @xmath102 in dependence on order of indices @xmath103 , equations ( [ 4.17 ] ) can be rewritten as follows @xmath85% \hat{\xi}_1(z , t)&=&\phantom{+}% \bar{\alpha}_1\sum_{a=1}^n\hat{t}_{10}^{(a)}(t)\,\delta(z_a(t)-z)\,% \hat{\xi}_3(z , t)\;-\;% \bar{\alpha}_2\sum_{a=1}^n\hat{t}_{xy}^{(a)}(t)\,\delta(z_a(t)-z)\,% \hat{\xi}_2(z , t)% \nonumber\\% \left[\frac{\partial}{\partial z}+\frac{1}{c}% \frac{\partial}{\partial t}\right]% \hat{\xi}_2(z , t)&=&\phantom{+}% \bar{\alpha}_2\sum_{a=1}^n\hat{t}_{xy}^{(a)}(t)\,\delta(z_a(t)-z)\,% \hat{\xi}_1(z , t)\;-\;% \bar{\alpha}_2\sum_{a=1}^n\hat{t}_{\xi\eta}^{(a)}(t)\,\delta(z_a(t)-z)\,% \hat{\xi}_3(z , t)% \nonumber\\% \left[\frac{\partial}{\partial z}+\frac{1}{c}% \frac{\partial}{\partial t}\right]% \hat{\xi}_3(z , t)&=&% -\bar{\alpha}_1\sum_{a=1}^n\hat{t}_{10}^{(a)}(t)\,\delta(z_a(t)-z)\,% \hat{\xi}_1(z ,
t)\;+\;% \bar{\alpha}_2\sum_{a=1}^n\hat{t}_{\xi\eta}^{(a)}(t)\,\delta(z_a(t)-z)\,% \hat{\xi}_2(z , t)% \label{5.2}%\end{aligned}\ ] ] where we introduced the dimensionless polarizabilities @xmath104these equations are valid for any type of initial conditions and can be simplified for excitation geometry described in the preamble to this section .
the coherent forward scattering of light linearly polarized along the @xmath21 direction does not modify the average angular momentum orientation of the atomic ensemble .
transmitted light also preserves its mean 100% linear polarization .
this means that in equations ( [ 5.2 ] ) only @xmath105 and @xmath69 have non - zero expectation values .
other operators exist only as fluctuating quantum variables .
it is possible to linearize these equations by substituting @xmath105 and @xmath69 by their average values and leaving in the right - hand side only the linearized contribution over quantum fluctuations .
then the evolution of the stokes operators is given by @xmath85% \hat{\xi}_1(z , t)&\approx&% -\kappa_2\,\hat{\xi}_2(z , t)\;+\;% \bar{\alpha}_1\,\bar{\xi}_3\,% \sum_{a=1}^n\,\hat{t}_{10}^{(a)}(t)\,\delta(z_a(t)-z)\,% \nonumber\\% \left[\frac{\partial}{\partial z}+\frac{1}{c}% \frac{\partial}{\partial t}\right]% \hat{\xi}_2(z , t)&\approx&\phantom{+}% \kappa_2\,\hat{\xi}_1(z , t)\;-\;% \bar{\alpha}_2\,\bar{\xi}_3\,% \sum_{a=1}^n\,\hat{t}_{\xi\eta}^{(a)}(t)\,\delta(z_a(t)-z)\,% \nonumber\\% \left[\frac{\partial}{\partial z}+\frac{1}{c}% \frac{\partial}{\partial
t}\right]% \hat{\xi}_3(z , t)&\approx&\phantom{+}0% \label{5.3}%\end{aligned}\ ] ] where @xmath106 is the average value of the corresponding stokes component , which is approximately unchanged for the light beam propagating through the sample .
the third line in eqs.([5.3 ] ) just indicates this circumstance .
the coupling parameter @xmath107is responsible for birefringence effects , i.e. for unitary transformation of linear polarization ( defined in respect to @xmath108 axes ) to circular polarization and vice versa . here
@xmath109 is the averaged and approximately unchanged value of the alignment of @xmath15-th atom . for any atom
, whose angular momentum @xmath29 is oriented along the @xmath21 direction , the alignment term is given by @xmath110^{1/2}}{2[2(j_0 + 1)(2j_0 + 1)(2j_0 + 3)]^{1/2}}\;% \equiv\;\bar{t}_{xy}% \label{5.5}%\]]the microscopic structure of @xmath111 can be averaged over a small mesoscopic interval of @xmath112 and substituted into eq.([5.3 ] ) as @xmath113 , where @xmath114 is the linear density of atoms ( i. e. the number of atoms per unit length ) . from the classical electrodynamics point of view , the mesoscopically averaged product @xmath115 is the difference between refraction indices for the polarizations along @xmath21 and @xmath22 directions .
the most important terms in equations ( [ 5.3 ] ) are the last ones on the right hand sides .
these terms show how any quantum state , originally encoded in the spin fluctuations of the atomic subsystem , can be mapped onto the polarization state of the light subsystem .
thus these terms are responsible for the quantum information processing in light - atoms interaction .
consider the dynamics of the orientation vector of an @xmath15-th atom . applying the commutation rule for irreducible tensor operators
@xmath116&=&% \delta_{ab}\,[(2k+1)(2k'+1)]^{1/2}\,% \sum_{k''}% \nonumber\\% & & \times [ 1-(-)^{k+k'+k''}]\,% \left\{\begin{array}{ccc}% k & k ' & k''\\ j_0 & j_0 & j_0% \end{array}\right\}\,% \nonumber\\% & & \times\,(-)^{2j_0+k''}\,c^{k''q''}_{kq\ , k'q'}\,\hat{t}_{k''q''}^{(a)}% \label{5.6}%\end{aligned}\]]see ref.@xcite , to the right hand side of equations ( [ 4.14 ] ) we obtain @xmath117\;-\;% \bar{\alpha}_2\,(-)^{2j_0}\sqrt{10}\,% \left\{\begin{array}{ccc}% 2 & 1 & 2\\ j_0 & j_0 & j_0% \end{array}\right\}\,% \hat{\xi}_3(z_a(t),t)\,\hat{t}_{\xi\eta}^{(a)}(t)% \nonumber\\% & & + \;\bar{\alpha}_2\,(-)^{2j_0}\sqrt{10}\,% \left\{\begin{array}{ccc}% 2 & 1 & 2\\ j_0 & j_0 & j_0% \end{array}\right\}\,% \hat{\xi}_1(z_a(t),t)\,\hat{t}_{xy}^{(a)}(t),% \nonumber\\% \nonumber\\% \dot{\hat{t}}{}_{1\pm 1}^{(a)}(t)&=&% \frac{i}{\hbar}\left[\hat{h}_0(t),\,\hat{t}_{1\pm 1}^{(a)}(t)\right]\;\pm\;% i\,\bar{\alpha}_2\,(-)^{2j_0}\frac{\sqrt{5}}{2}\,% \left\{\begin{array}{ccc}% 2 & 1 & 2\\ j_0 & j_0 & j_0% \end{array}\right\}\,% \hat{\xi}_0(z_a(t),t)\,\hat{t}_{2\pm 1}^{(a)}(t)% \nonumber\\% & & \mp\;i\,\bar{\alpha}_2\,(-)^{2j_0}\frac{\sqrt{5}}{2}\,% \left\{\begin{array}{ccc}% 2 & 1 & 2\\ j_0 & j_0 & j_0% \end{array}\right\}\,% \hat{\xi}_3(z_a(t),t)\,\hat{t}_{2\mp 1}^{(a)}(t)% \nonumber\\% & & + \;\bar{\alpha}_2\,(-)^{2j_0}\frac{\sqrt{5}}{2}\,% \left\{\begin{array}{ccc}% 2 & 1 & 2\\ j_0 & j_0 & j_0% \end{array}\right\}\,% \hat{\xi}_1(z_a(t),t)\,\hat{t}_{2\mp 1}^{(a)}(t)% \nonumber\\% & & \pm\;i\,\bar{\alpha}_1\,\frac{\sqrt{3}}% { 2[j_0(j_0 + 1)(2j_0 + 1)]^{1/2}}\;% \hat{\xi}_2(z_a(t),t)\,\hat{t}_{1\pm 1}^{(a)}(t)% \label{5.7}%\end{aligned}\ ] ] as one can see these equations are not closed , since the right hand sides are expressed in terms of alignment components .
moreover the higher rank multipoles drive the dynamics of the alignment components .
thus to introduce the system of closed heisenberg equations it is necessary to consider the coupled dynamics of all irreducible components defined for each atom .
however the dynamics of the atomic subsystem can be approximated by the dynamics of orientation components only as far as we are restricted to the discussion of the excitation regime described in the preamble to this section .
as shown in the appendix [ a ] equations ( [ 5.7 ] ) can be transformed to the set of nonlinear equations , where the right - hand sides are expressed in terms of the operators of the angular momentum . in a special , but the most important for us , regime of small fluctuations
such transformed equations can be linearized and simplified to the equations describing the dynamics of the vector of the angular momentum in a closed form .
let us define the vector of the collective angular momentum of ensemble as @xmath118then starting from equations ( [ a.7 ] ) , written for any @xmath15-th atom , and making the sum over all atoms of the ensemble , we arrive at the following equations governing the dynamics of the collective angular momentum @xmath119where parameters @xmath120 and @xmath121 are defined by expressions ( [ a.8 ] ) and ( [ a.9 ] ) respectively and @xmath122 is given by ( [ 5.5 ] ) .
the last source - type terms on the right hand side of these equations are responsible for the mapping of any quantum state , originally prepared in the stokes fluctuations of the field subsystem , onto long lived atomic spin subsystem . equations ( [ 5.3 ] ) and ( [ 5.9 ] ) , considered together , approximate the dynamical evolution of coupled collective variables of light and atomic subsystems .
but as follows from the structure of these equations they are still not closed because the coupling terms _ are not expressed by collective variables_. as we see from ( [ 5.3 ] ) only the atoms currently located before the wavefront contribute into formation of the fluctuating stokes variables . in turn , such spatially dependent stokes fluctuations , which stay actually unknown , drive the dynamics of atomic collective angular momentum via source terms in ( [ 5.9 ] ) .
however , as we show below , under certain simplifying assumptions , these equations can be further transformed into the system of closed equations describing the wave - type spatial and temporal distribution of the collective heisenberg operators of the atomic and field subsystems .
if atoms are slowly drifting in space and during the interaction with a short probe light pulse each atom preserves its location inside the area much less than the length scale comparable with @xmath123 or with the sample size , equations ( [ 5.3 ] ) and ( [ 5.9 ] ) can be transformed into a closed form .
let us note here that if in the experiment the duration of a probe pulse is chosen less than a few microseconds such a condition is normally fulfilled even in case of atoms at room temperature .
for the case of cold trapped atoms this assumption is consistent with the pulse duration up to a second .
then the equations of motion can be rewritten for any mesoscopic layer of the sample , which gives only a small increment to the heisenberg operators but contains a large number of atoms .
the atoms do not leave the layer during the interaction time and cooperatively interact with the electromagnetic field .
if this layer has a length of @xmath112 , we can introduce the averaged stokes operators @xmath71 along this layer . as a next step , instead of total angular momentum , given by eq.([5.8 ] )
we can define its mesoscopic spatial distribution as follows @xmath124where the sum over @xmath15 is extended only over the atoms located inside the layer .
then the total angular momentum of the ensemble is expressed as @xmath125where @xmath126 and @xmath64 is the length of the sample . the above assumptions lead us to the following set of closed and coupled equations for the mesoscopically averaged spatial distributions of the field and atomic variables @xmath85% \hat{\xi}_1(z , t)&=&% -\kappa_2\,\hat{\xi}_2(z , t)\;+\;% 2\beta\,\bar{\xi}_3\,% \hat{{\cal j}}_z(z , t)% \nonumber\\% \left[\frac{\partial}{\partial z}+\frac{1}{c}% \frac{\partial}{\partial t}\right]% \hat{\xi}_2(z , t)&=&% \phantom{+}\kappa_2\,\hat{\xi}_1(z , t)\;-\;% 2\epsilon\,\bar{\xi}_3\,% \hat{{\cal j}}_y(z , t)% \nonumber\\% \frac{\partial}{\partial t}\hat{{\cal j}}_z(z , t)&=&% \phantom{+}\omega\,\hat{{\cal j}}_y(z , t)\;-\;% \theta_y\bar{{\cal j}}_x\,% \hat{\xi}_1(z , t)% \nonumber\\% \frac{\partial}{\partial t}\hat{{\cal j}}_y(z , t)&=&% -\omega\,\hat{{\cal j}}_z(z , t)\;+\;% \theta_z\bar{{\cal j}}_x\,% \hat{\xi}_2(z , t)% \nonumber\\% & & \label{5.12}%\end{aligned}\]]where we used the same notation for the averaged stokes variables as for their microscopic origins and denoted the mesoscopic spatial distributions of angular momentum components as @xmath127 , with @xmath128 .
it is taken into account that @xmath129 .
equations ( [ 5.12 ] ) should be accompanied by corresponding initial and boundary conditions , which are given by @xmath130the solution of these equations presented in the next section shows how the swapping of quantum fluctuations between light and spin subsystems takes place during the interaction process .
several new parameters appear in equations ( [ 5.12 ] ) .
firstly , by @xmath131 we denote the frequency @xmath132 of the regular precession caused by the external magnetic field as well as by the light - induced shift of the zeeman sublevels .
secondly , there are two new parameters in the first pair of equations describing the transformation of the stokes variables .
the angle @xmath133 @xmath134^{1/2}}\,% \bar{\alpha}_1% \label{5.14}%\]]is the angle of faraday - type rotation of the polarization plane of the propagating light per one spin flip in the ensemble in @xmath6-direction .
the parameter @xmath135 @xmath136^{1/2}}% { 2[2j_0(j_0 + 1)(2j_0 + 1)(2j_0 + 3)]^{1/2}}\,\bar{\alpha}_2% \label{5.15}%\]]is the ellipticity induced in the propagating light by the atomic sample per one spin flip in @xmath22-direction , see figure [ fig.1 ] .
thirdly , there are two angles in the second pair of equations describing the dynamics of the spatial distribution of the atomic angular momenta .
angle @xmath137 is the rotation angle of the local collective angular momentum , originally oriented along @xmath21-axis , around @xmath22-axis per one photon propagating trough the sample in either @xmath72-type or @xmath73-type linear polarization . in turn ,
angle @xmath138 is the rotation angle of the local angular momentum around @xmath6-axis per one photon propagating through the sample in either right - hand - type or left - hand - type circular polarization .
for pedagogical purposes , we consider at first a special example of optical excitation in the far - off - resonance wing of @xmath139 or @xmath140 optical transitions and will discuss the general case after that . in the case of @xmath142 or @xmath140 transition equations ( [ 5.12 ] ) are simplified to the following form @xmath143\hat{\xi}_1(z , t)&=&% \phantom{+}2\beta\,\bar{\xi}_3\,% \hat{{\cal j}}_z(z , t)% \nonumber\\% \left[\frac{\partial}{\partial z}\,+\,\frac{1}{c}% \frac{\partial}{\partial t}\right]\hat{\xi}_2(z , t)&=&\phantom{+}0% \nonumber\\% \frac{\partial}{\partial t}\hat{{\cal j}}_z(z , t)&=&% \phantom{+}\omega_0\,\hat{{\cal j}}_y(z , t)% \nonumber\\% \frac{\partial}{\partial t}\hat{{\cal j}}_y(z , t)&=&% -\omega_0\,\hat{{\cal j}}_z(z , t)\;+\;% \beta\bar{{\cal j}}_x\,% \hat{\xi}_2(z , t)% \nonumber\\% & & \label{6.1}%\end{aligned}\]]where it was taken into account that @xmath144 and therefore @xmath145 , @xmath146 , @xmath147 . because of the second line in the system ( [ 6.1 ] ) , one has @xmath148 .
the straightforward solution for the distribution of the angular momentum components leads to @xmath149 and the stokes component @xmath71 is given by @xmath150\;% \hat{{\cal j}}_z^{in}(z')% \;+\;2\beta\,\bar{\xi}_3\,%
\int_0^z \!dz'\,\sin\omega_0[t-(z - z')/c]\;% \hat{{\cal j}}_y^{in}(z')% \nonumber\\% & & + \;2\beta^2\,\bar{\xi}_3\bar{{\cal j}}_x% \int_0^z \!dz'\!\int_0^{t-(z - z')/c}\!dt'\sin\omega_0[t - t'-(z - z')/c]\;% \hat{\xi}_2^{in}(t'-z'/c)% \label{6.3}%\end{aligned}\ ] ] this solution explains the basic idea of entanglement of the collective quantum states of atomic and field subsystems .
after the interaction cycle the quantum state of light described by the @xmath151 component is mapped onto the atomic angular momentum components because of the last terms in eqs.([6.2 ] ) . in turn , the information written in either @xmath152 or @xmath153 components of the angular momentum is mapped onto the @xmath154 stokes component of light , because of the second and third terms in eq.([6.3 ] ) . moreover , due to atom - field interaction there is a partial mapping of the @xmath151 component onto @xmath154 component because of the last term in eq.([6.3 ] ) .
the retardation effects , which are clearly visible in the derived solution are mainly important in order to preserve the proper commutation relations for the whole set of the heisenberg operators involved in the process .
the initial zero moment of time is a conventional and arbitrary chosen parameter to coordinate schrdinger and heisenberg pictures .
physically the interaction cycle starts only when the wavefront of the probe radiation crosses @xmath155 point .
we will take this time as the initial moment @xmath156 . in the heisenberg formalism
this means that any expectations values of @xmath157 and @xmath158 operators or their products are equal to zero for time @xmath159 .
since the goal is to evaluate the expectation values of outgoing operators it is acceptable to substitute the lower zero limit by @xmath160 in ( [ 6.2 ] ) and by @xmath161 in the internal integral ( [ 6.3 ] )
. then the derived solution can be further applied at any spatial point @xmath6 only for time @xmath162 , which is in accordance with physical manifestation of retardation effects . in reality
, we are most interested in a correct description of the behavior of the modes around the carrier frequency @xmath4 , see definition ( [ 4.15 ] ) . for the modes distributed in the spectral domain much narrower than @xmath163 , where @xmath64 is the length of the sample , the retardation effects become negligible . in this case
the above solution is simplified and can be written in terms of transverse components of the collective angular momentum @xmath164 and for the output stokes operator @xmath165 @xmath166which coincides , in principle , with the solution obtained earlier in refs.@xcite .
the input - output transformations expressed by ( [ 6.4 ] ) and ( [ 6.5 ] ) have a simple logical structure and therefore are attractive for possible applications to the quantum information tasks .
however , they are only valid for @xmath167 isolated transitions , and it is rather difficult to find a practical example of such a transition . however , the same simple input - output relations are effectively realized in the case of alkali atoms excited in the far wing of @xmath168 or @xmath169 lines . in this case
the contributions from the components of the upper hyperfine multiplet to the interaction add up in such a way that the alignment effects become unimportant .
for a transition from the ground hyperfine state @xmath170 to all hyperfine states @xmath171 we straightforwardly obtain after taking the sum over the excited states @xmath172hence the alignment contribution vanishes for the ground state electronic angular momentum @xmath173 if the frequency detuning @xmath174 for each hyperfine transition @xmath175 becomes larger than the hyperfine splitting of the upper state . the crucial inequality @xmath176 ) . in spite of
the fact that for heavy alkali atoms the ratio of the hyperfine to fine splitting in the upper state @xmath177 is really small ( it is less than @xmath178 for @xmath179cs ) , which makes it possible to neglect the alignment contribution for large detunings , such an approximation is not always easy to fulfill under typical experimental conditions .
this is because for a large frequency detuning one has to use pulses with a very large photon number in order to make the atom - field interaction strong enough .
large photon numbers are often inconvenient due to either limitations of detectors or due to a limited pulse duration , or both . therefore the frequency detuning @xmath180
is frequently chosen to be comparable with @xmath181 .
a typical detuning chosen in the experiment @xcite , is of the order of @xmath182 with the hyperfine splitting in @xmath183 of @xmath184 .
this is one of the motivations for the deeper analysis of equations ( [ 5.12 ] ) in their general form
. equations ( [ 5.12 ] ) can be solved in the general case , which we do first ignoring the retardation effects .
this is the most important case for practical applications to quantum information processing . without retardation
the system of coupled equations ( [ 5.12 ] ) can be rewritten as follows @xmath185this is a set of differential equations with constant coefficients and it is accompanied by initial and boundary conditions ( [ 5.13 ] ) .
the solution can be presented in the form of an integral response on these conditions . for the stokes
operators it can be written as @xmath186with @xmath187 . for the operators of the spatial distributions of the angular momentum components the solution is @xmath188with @xmath189 .
the solution can be formally extended over the space - time region @xmath190 and @xmath191 . as shown in the appendix [ b ] , the kernels of the integral transformations in ( [ 6.8 ] ) and ( [ 6.9 ] ) can be expressed via the corresponding laplace images @xmath192where @xmath193 is any of the matrix functions @xmath194 in equations ( [ 6.8 ] ) and ( [ 6.9 ] ) and @xmath195 is its laplace image .
the limits of integration @xmath196 and @xmath197 are chosen as arbitrary real values warranting the existence of laplace images for the spatial and temporal transforms respectively . in the appendix [ b ] the images for all the matrices
@xmath198 are calculated explicitly .
evaluation of the integrals ( [ 6.10 ] ) can be done numerically for any sample when all the external parameters are defined .
let us briefly explain how the retardation effects can be taken into account . to do this
the retardation time @xmath199 should be introduced as an independent variable for each spatial point @xmath6 at a certain moment of time @xmath19 .
then the basic system ( [ 5.12 ] ) , where all the operators are considered now as functions of @xmath6 and @xmath200 , can be converted to the form ( [ 6.7 ] ) but with the derivative over retardation time on the left hand side .
the initial conditions should be also modified and written for the initial retardation time @xmath201 , which corresponds now to spatially dependent moments in the laboratory time @xmath202 . as before we will assume that the wavefront of the probe radiation crosses @xmath155 point at zero time . in this case
we need to know the cooperative atom - field dynamics at any spatial point @xmath6 only for time @xmath162 .
then we may ignore the interaction hamiltonian for time @xmath203 and chose the initial conditions in ( [ 5.13 ] ) in the following form @xmath204there are no changes in the boundary conditions in ( [ 5.13 ] ) , since at @xmath155 the retardation time @xmath200 coincides with @xmath19 .
thus in the most general case the solution ( [ 6.8])-([6.10 ] ) should be rewritten in terms of the retardation time and should depend on @xmath205 and @xmath206 as on initial quantum fluctuations of the atomic angular momenta .
finally the retardation effects give us the following correction to the spatial and temporal dynamics of the field variables @xmath207 and of the atomic variables @xmath208 these expressions can be applied for calculation of expectation values of any products of operators at a spatial point @xmath6 for the time @xmath162 . an important feature of solutions ( [ 6.12 ] ) and ( [ 6.13 ] ) is their wave nature . as follows from the laplace transform carried out in the appendix [ b ] the eigenmodes of spatial and temporal distributions of atoms - field collective fluctuations
are associated with the poles of determinant @xmath209 , where @xmath210 is defined by ( [ b.3 ] ) . for atoms with spin @xmath211
this condition leads to @xmath212 and @xmath213 . in the absence of the magnetic field @xmath214
there are only two collective modes associated with the total number of atoms and with the total number of scattering photons which undergo polarization sensitive interaction in the process of coherent forward scattering .
this makes it possible to further simplify the effective hamiltonian and to discuss it in terms of integrated collective variables of the canonical type @xcite .
however in a general case , due to presence of alignment associated effects , the condition @xmath209 gives coupled roots @xmath215 , i.e the wave - type modes which can survive in the sample and be excited by either time dependent fluctuations of the field stokes components or by spatial fluctuations of atomic spins .
such quantum superposition of the field and material waves can be understood as a polariton - type spin wave generated in the medium .
the multimode structure of such waves does not allow entanglement only between integrated collective variables . instead , in general the entanglement
is distributed in all polariton modes . from the quantum information point of view
this means that in the case of atoms with the ground state angular momentum equal or larger than @xmath216 a possible entanglement resource hidden in the polariton modes allows for multimode entanglement rather than a single mode entanglement available for spin one half systems .
the quantum information then will be mapped on spatially dependent correlations of an atomic standing spin wave . the following remark concerning the terminology is due here .
let us point out the principle difference between the spin - wave dynamics described by eqs.(5.13 ) and the collective spin behavior in the spin - organized cold fermion gas , which was discussed years ago , see @xcite . in our case
the cooperative dynamics of atomic spins is driven by the radiation - type interaction whereas in the cold fermionic systems it is driven by the static interparticle interaction via the longitudinal electric field . in this section
we concentrate on calculations relevant for the existing experimental example @xcite carried out with ensemble(s ) of @xmath179cs atoms .
we first calculate the degree of spin squeezing ignoring the alignment , i.e. , following the model adopted in @xcite and describing the atomic ground state by the spin ( orientation ) only .
we then present the numerical results including the alignment and show how this affects spin squeezing of atoms . in the experiment @xcite the entanglement ( two mode squeezing )
was generated between two spatially separated via a faraday - type detection of light . for pedagogical reasons we will make numerical simulations for the case of single mode spin squeezing , which makes no difference for the present discussion .
we will completely ignore the retardation effects and consider the spin dynamics without external magnetic field .
in the case of cesium atoms the alignment contribution can be suppressed if the frequency detuning of the probe light from the atomic resonance is much larger than the hyperfine splitting in the upper state , as described in the end of section [ s6.a ] . if the alignment is ignored the basic equations stem from the expressions ( [ 6.4 ] ) and ( [ 6.5 ] ) , which in the absence of magnetic field read @xmath217 the input - output transformations ( [ 6.14 ] ) show the entangling mechanism of the atoms - field variables in the process of coherent forward scattering .
if the number of atoms and photons is large enough such as @xmath218 and @xmath219 the output quantum fluctuations become strongly entangled . indeed , in this case
the role of @xmath220 and @xmath157 terms on the right hand side of ( [ 6.14 ] ) becomes negligible if these fluctuations have been originally poissonian . after the interaction
there would be strong correlations between fluctuations of @xmath221 and @xmath222 as well as between @xmath223 and @xmath224 .
if the observable @xmath225 is measured by a balanced faraday detector there will be no longer standard quantum uncertainty in @xmath226 , whose original ( at @xmath227 ) poissonian fluctuations will be suppressed by a factor @xmath219 .
the collective spin state will become squeezed and will accumulate its quantum uncertainty in @xmath228 fluctuation .
it is important to recognize that even in such an ideal scheme there is a restriction on the number of participating atoms and photons because of accumulation of events of the incoherent scattering .
it is clear that in each event of the incoherent scattering one photon and one atom of the ensemble cancel out from entangling process .
these losses can be neglected if the number of scattering events is much less than the total numbers of atoms @xmath229 and incoming photons @xmath230 .
this can be written as two inequalities @xmath231where @xmath232 is the cross section for off - resonant incoherent scattering in full solid angle , and @xmath18 is the area of the light beam , which coincides with the area of atomic cloud .
both inequalities lead to similar restrictions on the number of participating atoms and photons @xmath233 in case of photons this inequality can be also understood as a restriction on the whole interaction time @xmath234 ( the probe pulse duration ) , since @xmath235 .
since both type of the losses are undesirable one can assume that @xmath236 to exclude any preference for atoms or photons .
the number of scattered photons can be expressed as @xmath237where by small parameter @xmath73 we denoted the relative number of atoms and photons lost as a result of incoherent scattering , see the first line of ( [ 6.15 ] ) .
a detailed analysis of the role of spontaneous emission is presented in @xcite .
the scheme of energy levels of @xmath179cs is shown in figure [ fig.2 ] . skipping technical details of calculation of off - resonant cross section @xmath232 ( on an atom in the zeeman sublevel @xmath238 of its ground state ) and of estimation of the elementary faraday angle @xmath133 , we present the final result .
the square variance of any output stokes component @xmath239 , where @xmath187 , is given by @xmath240 ^ 2\right\rangle\;=\;% \bar{\xi}_3t\,\left[1\,+\,\xi_i(j)\right]% \label{6.18}%\]]where the mandel parameter @xmath241 , considered as function of the total collective angular momentum @xmath242 , shows the relative deviation from the shot - noise level .
this deviation is due to the variance of the atomic state mapped onto the output light .
for @xmath243 stokes component the original value of the mandel parameter @xmath244 transforms as @xmath245where the second line defines the dimensionless parameter @xmath246 responsible for coupling of the field and atomic subsystems described in terms of their canonical variable , see @xcite . for @xmath247 component
the mandel parameter preserves its magnitude , so that @xmath248 . in ( [ 6.19 ] ) and throughout we approximate the input fluctuations as markovian delta - correlated process in the lower frequency domain @xmath249and assume the poissonian ( coherent state ) square variance of the angular momentum @xmath250 for original coherent atomic spin state .
second term in ( [ 6.19 ] ) denotes the contribution of the faraday effect itself and function @xmath251 denotes the product @xmath252where @xmath253 is the rate of the spontaneous decay of the upper state and @xmath254 with @xmath255 are the frequency detunings of the probe light from each exciting hyperfine transition , see figure [ fig.2 ] .
-line of @xmath179cs : @xmath256 , @xmath257 , and @xmath258 are the frequency detunings of the probe light from hyperfine transitions participating in the process . ]
note that , as a consequence of neglecting the alignment effects , the total angular momentum ( or a number of atoms ) contributes in the output mandel parameter only in combination @xmath259 .
this parameter is the optical activity of the sample , i.e. the angle of faraday rotation of the planar wave for the ensemble perfectly oriented along the probe beam , see figure [ fig.1 ] .
this linearity ( see dashed - dotted lines in figures [ fig.3]-[fig.5 ] ) has been used as the benchmark for the determination of the projection noise level of atoms in all the work related to quantum state generation with atomic ensembles , see for example @xcite . as follows from the above discussion of the role of spontaneous emission , a reasonable choice of parameters for spin squeezing corresponds to @xmath260 . under typical conditions for the frequency detuning @xmath261 and
@xmath262 the optical activity can be @xmath263 , which can provide spin squeezing far below the standard quantum limit in ( [ 6.18 ] ) . in order to calculate the corrections to the variances of the output stokes components for a realistic alkali atom at a finite detuning the alignment associated effects have to be included in calculations .
this can be done only numerically by means of the transformations ( [ 6.8 ] ) . in figures
[ fig.3]-[fig.5 ] we plot the results of these calculations .
the variances of the stokes components @xmath243 and @xmath247 are shown as a function of the optical activity of the sample @xmath264 , calculated for different frequency offsets @xmath265 and for @xmath266 . in the figures
the solid curves represent the complete result within the described model , whereas dashed curves are computed only including the partial contribution coming from the first term in eq.([6.8 ] ) .
these curves show the excessive atomic spin fluctuations , mapped onto the stokes collective variable of the transmitted light , beyond the level of the transformed original light fluctuations and beyond the level of the spin ( orientation ) projection noise .
the incoming light is taken to be in a coherent state with poissonian statistics and with @xmath267 .
the enhancement of the light noise is caused by higher orders of the interaction process .
considered at certain value of @xmath259 the difference between the complete variance and the background level shows the efficiency of the faraday detector as a spin squeezer .
the scheme works the better the higher this difference is .
however , because of the alignment the squeezing has to be considered not for a collective angular momentum but for a more complex type of a spatial spin mode , which is defined by the input - output transformations ( [ 6.8 ] ) and ( [ 6.9 ] ) . squeezing and entanglement for such spatial modes will be considered elsewhere .
( upper plot ) and @xmath268 ( lower plot ) as a function of optical activity of the sample for the frequency offset @xmath269 .
the dotted curve indicates the original shot - noise level , the dashed curve is the atomic signal coming from the first term in ( [ 6.8 ] ) , the dashed - dotted linear dependence is the faraday approximation ( [ 6.19 ] ) , the solid curve is the complete output variance . ] for the frequency offset @xmath270 . ] for the frequency offset @xmath271 . ]
it is instructive to compare the result of numerical calculations with the approximation ignoring the alignment effects .
the dashed - dotted linear dependencies in figures [ fig.3]-[fig.5 ] reproduces the ideal result described by eq.([6.19 ] ) . as we see , for small values of @xmath259 the faraday approximation fits the exact solution with rather good accuracy and the deviation of @xmath272 from zero is quite small .
this means that for the sample characterized by rather small optical activity the spin one - half approximation for the multi - level cesium atom is self - consistent and potentially good for describing the real experimental situation .
hence for such a sample the squeezed spin standing wave can be approximated by the collective angular momentum .
let us point out here that in the existing experiment @xcite the optical activity was @xmath273 and the alignment correction was not so important .
however , as clearly seen for large @xmath259 the difference between the input - output transformations in their general form and approximation ( [ 6.14 ] ) becomes quite important .
we have considered the quantum theory of coherent forward scattering of light by an ensemble of multi - level atoms polarized in their angular momenta . as a result of such a scattering process
the quantum states of the field and atomic subsystems are transformed into an entangled state . in our discussion of the process we followed the effective hamiltonian approach , which in the semiclassical form is normally applied for studying optical pumping processes after adiabatic elimination of the excited state . compared to earlier studies of entanglement of light and atomic ensembles , we have derived the effective hamiltonian in a more general form for the atoms with an arbitrary total magnetic momentum . towards this end , we have found the physical conditions under which the analysis can be simplified by introducing a finite number of collective variables for light and atoms .
we showed that under certain conditions the cooperative atoms - field dynamics can be properly described by the wave - type coupled equations for the space - time evolution of the collective heisenberg operators of the field and atomic subsystems . in these equations an infinite set of atom - field operators is truncated via introduction of slowly varying collective modes .
the coupled equations for the time evolution of spatially dependent operators of the stokes components of light and of the macroscopic fluctuations of the collective atomic angular momentum are written in a closed form . in the general case
the coupled dynamics of the atom and field operators manifests itself in a spin polariton wave created in the sample .
the fluctuating components of atoms and field become strongly entangled in the polariton wave in space and time .
such spin polariton waves initiated by radiative forces are different from the collective spin dynamics existing in the spin - polarized quantum gas @xcite .
they are also different from the polariton modes discussed in @xcite where polaritons of the combined atom - light state are introduced . in our case
the quantum entanglement arises from the interaction between the internal collective polarization degrees of freedom of light and atomic subsystems .
in particular our analysis yields the input - output transformations for the heisenberg operators of the collective variables of light and atomic spins after the whole interaction cycle .
our results suggest that a successful approach of using off - resonant light - atomic ensemble interaction for quantum information processing can become even more fruitful with the use of a multi - mode type of entanglement provided by spin polariton atomic variables .
numerical simulations demonstrate the importance of the developed formalism for application to a realistic experimental situation .
we considered a well known example , when the faraday rotation is used as a non - demolishing measurement of transverse fluctuations of the collective atomic spin and can be utilized as a physical mechanism for the spin squeezing in an ensemble of spin polarized atoms .
we tested the validity of the spin one - half approximation for a realistic cesium atom , which is normally used to describe the interaction with a far off - resonant probe light around @xmath168 or @xmath169 transitions of alkali atoms .
as shown , for small values of optical activity this approximation is self - consistent and deviates negligibly from the calculations based on a general solution .
however for the samples with high optical activity , where quantum correlations become strong , there is an important quantitative as well as qualitative difference between the general solution and the model of spin one - half atoms .
financial support for this work was provided by intas ( grant info 00 - 479 ) , by danish national research foundation , by the eu grant covaqial , and by the north atlantic treaty organization ( pst - clg-978468 ) .
d.v.k . would like to acknowledge financial support from the delzell foundation , inc .
the unperturbed electric field operator in the origin of the coordinate frame coupled with a scattering atom , which drifts with velocity @xmath274 , is given by @xmath275% \nonumber \\% & = & \hat{\mathbf{e}}_{0}^{(-)}(t)\;+\;\hat{\mathbf{e}}_{0}^{(+)}(t)% \label{2.1}%\end{aligned}\]]where @xmath276 , @xmath277 are respectively the operators of creation and annihilation of the photon with wavevector @xmath278 and in the polarization state @xmath279 .
@xmath280 is the quantization volume .
the second line in ( [ 2.1 ] ) defines negative and positive frequency components of the electric field .
the dipole - type interaction operator of the atom with the electric field is given by @xmath281where the operator for the atomic dipole moment @xmath282 and for the electric field are defined in the interaction representation . based on a perturbation theoretic approach
, the exact solution for the heisenberg operator can be written as the following expansion @xmath283which can be also written for positive @xmath284 and negative @xmath285 frequency components .
we assume that the wave function of the joint atom - field system describes the combined state where an atom occupies the ground state and the electric field is in a weak quasi - coherent state non - saturating the atomic transition .
then the correction of the first order in ( [ 2.3 ] ) will disappear after averaging over the wave function .
thus the second order correction gives us the main contribution , since it is responsible for the scattering process .
the second order term in ( [ 2.3 ] ) can be written as follows @xmath286 \right]% \label{2.4}%\end{aligned}\]]as follows from this expression , in a complete dynamical description of the process there is a memory of initial conditions in the formal expansion of perturbation theory . however , for non - saturating fields this solution can be spread out over the time @xmath287 , where @xmath288 is the natural radiative relaxation rate of the upper state .
but in this case it is necessary to take into account all the radiative correction for the retarded and advanced green functions of the decaying upper atomic state .
this can be done by introducing the natural decay law into the time behavior of these functions .
then the integral ( [ 2.4 ] ) loses its dependence on the lower limit as on initial time coordinate .
let us also point out that there is only conventional choice of initial time @xmath289 to coordinate the heisenberg and schrdinger representations .
the real physical conditions can be arranged as the wave front of probe radiation ( expressed by expectation values of any products of freely heisenberg operators of the field ) could arrive to the interaction area at any time after @xmath290 .
thus we can always think that observation time @xmath19 satisfies inequality @xmath291 .
at the same time it is important to recognize that by including the radiation decay into the green functions of the upper atomic states we average their time evolution and ignore any random fast variation of the field and atomic operators associated with high frequencies of the field continuum .
the integral ( [ 2.4 ] ) gives us the solution for the electric field in the origin of the frame coupled with a moving atom .
but in the zero order of relativistic effects , when only retardation effects in the radiation zone have to be taken into consideration , this solution coincides with the electric field in the laboratory frame at the point of atom location as well as in the small vicinity of this point .
then we can obtain the solution for any point in the laboratory frame by using the propagation law in free space . by this procedure one
obtains the following expansion for the positive frequency component of the electric field operator in the radiation zone ( @xmath292 ) of the scattering atom @xmath293where @xmath294 and @xmath295% \label{2.7}%\end{aligned}\ ] ] and the negative frequency component is given by the hermitian conjugation : @xmath296 . here
the origin of the frame is chosen in the location of the atom , which is assumed to be unchanged during the light propagation time @xmath297 .
the scattered light frequency @xmath298 is defined here via the input frequency @xmath299 , and the raman shift @xmath300 for atomic transition @xmath301the wave vector @xmath302 , but in the right hand side of ( [ 2.8 ] ) the doppler correction is neglected and it is assumed that @xmath303 .
the transition dipole moments in ( [ 2.7 ] ) are defined in the schrdinger representation and its transverse component is given by @xmath304the sum over @xmath305 is expanded over all possible excited transitions characterized by natural linewidths @xmath306 , but as a practical matter , the sum can be restricted to the most significant resonance transitions and the frequency @xmath307 can be associated with the frequency of the incident mode . in spite both the terms in ( [ 2.6 ] ) are well known in the scattering theory , see @xcite , we claim that only the latter term , constituted with rotating wave approximation , should be left in aiming the following generalization to multiple scattering process .
indeed , the fist term contributed in ( [ 2.6 ] ) reveals the quantum mechanical phenomenon than the scattered photon is created in advance than incoming photon was annihilated .
this process has very small , and in fact negligible , amplitude and should be ignored if one stays with semiclassical - type understanding of multiple scattering process as sequence of successive scattering events .
moreover , as shown in @xcite , the electric field transformation in the presence of one scatterer in form ( [ 2.5])-([2.7 ] ) preserves commutation relation , i. e. being unitary transformation only in rotating wave approximation . in the rotating wave approximation the perturbation theory solution ( [ 2.5])-([2.7 ] ) satisfies the following equation @xmath308where @xmath309and in turn @xmath310 is the operator of transverse component of atomic polarization responding on external field .
frequency @xmath311 is given by eq.([2.8 ] ) , but wavevector @xmath312 is independent variable here . in above discussion
the interaction time @xmath19 was assumed to be short enough for validity of the perturbation theory approach .
therefore in the derived equations the dyadic - type operators @xmath313 of the low atomic state can be selected as referred to interaction representation @xmath314but this operators only slightly differs from their heisenberg analogs during the short interaction time .
therefore the evolution of the electric field operator can be extended up to arbitrary time @xmath19 , if the final equations of this section are modified as follows .
all the dyadic - type operators should be changed by the corresponding heisenberg operators with keeping the complete dynamic evolution up to moment @xmath19 @xmath315 the drift of the atom in space can not be further ignored and instead of origin of the laboratory coordinate frame we should assume its actual location associated with its classical motion @xmath316 .
equation ( [ 2.10 ] ) stays valid and unchanged but atomic polarization ( [ 2.12 ] ) modifies to @xmath317 and @xmath318 . in substituting ( [ 2.15 ] ) into eq.([2.10 ] ) the time derivation should be done only for the fast oscillating components of the heisenberg operators and for the exponential factors .
the operators of irreducible components can be expressed in terms of the operators of atomic angular momenta . for any @xmath15-th atom of ensemble
the orientation vector is given by @xmath319^{1/2}}\,\hat{j}_q^{(a)}% \label{a.1}%\]]and the alignment tensor is given by @xmath320^{1/2}}\,% \nonumber\\% & & \times\sum_{qq'}\,c^{2q}_{1q\,1q'}\,% \left[\hat{j}_q^{(a)}\hat{j}_{q'}^{(a)}\,+\,% \hat{j}_{q'}^{(a)}\hat{j}_q^{(a)}\right.% \nonumber\\% & & \left.\phantom{c^{2q}_{1q\,1q'}}-\;% ( -)^q\delta_{q ,- q'}\frac{2}{3}j_0(j_0 + 1)\right]% \label{a.2}%\end{aligned}\ ] ] where @xmath321 are the cyclic components of the operator vector of angular momentum , which are defined by their cartesian components as follows @xmath322% \label{a.3}%\end{aligned}\]]see @xcite . by substituting subsequently ( [ a.1])-([a.3 ] ) into equations ( [ 5.7 ] ) the latter can be straightforwardly transformed to the set of nonlinear equations containing only the operators of atomic angular momenta @xmath323^{1/2}}\,% \left\{\begin{array}{ccc}% 2 & 1 & 2\\ j_0 & j_0 & j_0% \end{array}\right\}\,\hat{\xi}_3(z_a(t),t)\,% \left[\hat{j}_{x}^{(a)}(t)\hat{j}_{y}^{(a)}(t)\,+\,% \hat{j}_{y}^{(a)}(t)\hat{j}_{x}^{(a)}(t)\right]% \nonumber\\% & & + \;\bar{\alpha}_2\,(-)^{2j_0}\frac{5}{[(2j_0 - 1)(2j_0 + 3)]^{1/2}}\,% \left\{\begin{array}{ccc}% 2 & 1 & 2\\ j_0 & j_0 & j_0% \end{array}\right\}\,% \hat{\xi}_1(z_a(t),t)\,% \left[\hat{j}_{x}^{(a)2}(t)\,-\,% \hat{j}_{y}^{(a)2}(t)\right],% \nonumber\\% \nonumber\\% \dot{\hat{j}}{}_{y}^{(a)}(t)&=&% -\omega_0\,\hat{j}_{z}^{(a)}(t)\;+\;% \bar{\alpha}_2\,(-)^{2j_0}\frac{5}{2[(2j_0 - 1)(2j_0 + 3)]^{1/2}}\,% \left\{\begin{array}{ccc}% 2 & 1 & 2\\ j_0 & j_0 & j_0% \end{array}\right\}% \nonumber\\% & & \times\left[\hat{\xi}_0(z_a(t),t)\,+\,\hat{\xi}_3(z_a(t),t)\right]% \left[\hat{j}_{x}^{(a)}(t)\hat{j}_{z}^{(a)}(t)\,+\,% \hat{j}_{z}^{(a)}(t)\hat{j}_{x}^{(a)}(t)\right]% \nonumber\\% & & + \;\bar{\alpha}_2\,(-)^{2j_0}\frac{5}{2[(2j_0 - 1)(2j_0 + 3)]^{1/2}}\,% \left\{\begin{array}{ccc}% 2 & 1 & 2\\ j_0 & j_0 & j_0% \end{array}\right\}\,\hat{\xi}_1(z_a(t),t)\,% \left[\hat{j}_{y}^{(a)}(t)\,\hat{j}_{z}^{(a)}(t)\,+\,% \hat{j}_{z}^{(a)}(t)\,\hat{j}_{y}^{(a)}(t)\,\right]% \nonumber\\% & & + \;\bar{\alpha}_1\,\frac{\sqrt{3}}{2[j_0(j_0 + 1)(2j_0 + 1)]^{1/2}}\;% \hat{\xi}_2(z_a(t),t)\,\hat{j}_{x}^{(a)}(t),% \nonumber\\% \nonumber\\% \dot{\hat{j}}{}_{x}^{(a)}(t)&=&% -\;\bar{\alpha}_2\,(-)^{2j_0}\frac{5}{2[(2j_0 - 1)(2j_0 + 3)]^{1/2}}\,% \left\{\begin{array}{ccc}% 2 & 1 & 2\\ j_0 & j_0 & j_0% \end{array}\right\}% \nonumber\\% & & \times\left[\hat{\xi}_0(z_a(t),t)\,-\,\hat{\xi}_3(z_a(t),t)\right]% \left[\hat{j}_{y}^{(a)}(t)\hat{j}_{z}^{(a)}(t)\,+\,% \hat{j}_{z}^{(a)}(t)\hat{j}_{y}^{(a)}(t)\right]% \nonumber\\% & & -\;\bar{\alpha}_2\,(-)^{2j_0}\frac{5}{2[(2j_0 - 1)(2j_0 + 3)]^{1/2}}\,% \left\{\begin{array}{ccc}% 2 & 1 & 2\\ j_0 & j_0 & j_0% \end{array}\right\}\,\hat{\xi}_1(z_a(t),t)\,% \left[\hat{j}_{x}^{(a)}(t)\,\hat{j}_{z}^{(a)}(t)\,+\,% \hat{j}_{z}^{(a)}(t)\,\hat{j}_{x}^{(a)}(t)\,\right]% \nonumber\\% & & -\;\bar{\alpha}_1\,\frac{\sqrt{3}}{2[j_0(j_0 + 1)(2j_0 + 1)]^{1/2}}\;% \hat{\xi}_2(z_a(t),t)\,\hat{j}_{y}^{(a)}(t)% \label{a.4}%\end{aligned}\ ] ] in their general form these equations are quite complicated and not closed because of their nonlinear structure , but they can be simplified and linearized in the following assumptions .
the dynamics of operators @xmath324 is driven only by those terms , which , being averaged , have a quadratic scale over fluctuations of the field and atomic variables .
recall that @xmath325 .
physically this means that there is no coherent process demolishing the original spin orientation of single atom along @xmath21 direction .
so far we neglected any possibilities of incoherent scattering our analysis has to be restricted by assumption that in average @xmath326 .
moreover , since this observable has a maximal possible expectation value and it has no deviation from @xmath29 in the lower orders of weak external perturbations , it is allowed to approximate the heisenberg operator @xmath324 when it appears in linear or in the squared nonlinear form by its non - perturbed projector onto atomic wavefunction @xmath327but while substituting it in the operators products one has to follow the rule @xmath328 this is crucial point of linearizing procedure , it is expected that the left and right - hand side operators are coordinated if the spin subsystem is slightly disturbed and the time dynamics of off - diagonal projectors @xmath329 and @xmath330 is only taken into consideration .
then orientation dynamics , described by ( [ a.4 ] ) , can be approximated by the following set of linearized equations @xmath331where @xmath332^{1/2}}% { [ 2j_0(j_0 + 1)(2j_0 + 1)(2j_0 + 3)]^{1/2}}\;% \bar{\alpha}_2\,\bar{\xi}_0% \label{a.8}%\]]is the light induced shift between @xmath333 and @xmath334 sublevels .
average value of alignment component @xmath109 , is defined in the body of the paper by eq.([5.5 ] ) .
similarly we introduced here the orientation component @xmath335 , associated with average orientation of @xmath15-th atom , as @xmath336^{1/2}}% { [ ( j_0 + 1)(2j_0 + 1)]^{1/2}}\;\equiv\;\bar{t}_{x}% \label{a.9}%\]]here @xmath335 is the average orientation component of @xmath337-type defined in the frame with @xmath31-axis along magnetic field , which coincides with @xmath21-axis in our case .
this component is the same for all the atoms of ensemble . making the sum over all partial equations ( [ a.7 ] ) we come to equations ( [ 5.9 ] ) written for the collective vector of atomic angular momentum .
let us define the laplace images of the space - time dependent stokes components of the probe light and of the collective angular momentum of atoms @xmath338where parameters @xmath339 .
then the original system of differential equations ( [ 6.7 ] ) with initial and boundary conditions ( [ 5.13 ] ) can be transformed to the following system of linear algebraic equations for the laplace images @xmath340the determinant of this system is given by @xmath341and its solution can be easy found by means of the kramer s rules .
the solution is expressed as linear transformations of the laplace images of the boundary stokes operators for the field and of the initial angular momentum operators for the atoms @xmath342these transformations perform the laplace images of the integral transforms ( [ 6.8 ] ) and ( [ 6.9 ] ) introduced in the body of the paper .
to return into original space - time dependent representation it is necessary to evaluate the integrals ( [ 6.10 ] ) .
as we see from ( [ b.3])-([b.8 ] ) the laplace images of all the matrix elements have polynomial structure and the return transform could be found for any definite set of the external parameters , such as @xmath351 etc .. | quantum theoretical treatment of coherent forward scattering of light in a polarized atomic ensemble with an arbitrary angular momentum is developed .
we consider coherent forward scattering of a weak radiation field interacting with a realistic multi - level atomic transition .
based on the concept of an effective hamiltonian and on the heisenberg formalism , we discuss the coupled dynamics of the quantum fluctuations of the polarization stokes components of propagating light and of the collective spin fluctuations of the scattering atoms . we show that in the process of coherent forward scattering this dynamics can be described in terms of a polariton - type spin wave created in the atomic sample .
our work presents a general example of entangling process in the system of collective quantum states of light and atomic angular momenta , previously considered only for the case of spin @xmath0 atoms .
we use the developed general formalism to test the applicability of spin @xmath0 approximation for modelling the quantum non - demolishing measurement of atoms with a higher angular momentum . |
the lcao - scf @xcite method is generally employed for molecules , in which molecular wave functions taken to be linear combinations of atomic basis functions whose should possess the cusps condition at the nuclei @xcite and decay exponentially for large distances @xcite .
this approach leads to use , namely , slater - type orbitals @xcite , @xmath0 here , @xmath1 are complex or real spherical harmonics @xmath2 differs from the condon@xmath3shortley phases by sign factor @xmath4 @xcite , @xmath5 are gamma functions @xcite , @xmath6 are the principal , orbital , magnetic quantum numbers with , @xmath7 , @xmath8 , @xmath9 and @xmath10 stands for the integer part of @xmath11 , respectively , in one@xmath3 and two@xmath3electron multi@xmath3center molecular integrals . these integrals needs to be calculated in spectroscopic accuracy in order to meaningful discussions on basis - set expansion methods , born - oppenheimer energy , vibrational frequency calculations . the difficulty of finding analytically closed form relations , however , for molecular integrals have more than two - center referred to as _ the bottleneck of quantum chemistry _
@xcite , have been greatest obstacle since slater - type orbitals have no simple addition theorem ; relations for products of two slater - type orbitals centered on different positions not available in compact form @xcite .
the slater - type orbitals are obtained by simplification of laguerre functions in hydrogen@xmath3like orbitals @xcite by keeping only the term of the highest power of @xmath12 , for integer values of principal quantum number @xmath11 ( istos ) , where @xmath13 , @xmath14 and it has been proved that they provide extra flexibility for closer variational description of atoms and molecules by considering the values of @xmath11 in more general set of number , namely positive real numbers ( nstos ) , where @xmath7 .
the studies on the evaluation of molecular integrals , thus , are performed in two main group : those restrict the principal quantum number with integer values , which are practically used in nonrelativistic molecular electronic structure calculations @xcite and those free them from any specification but also reduce the area of applications only to investigation of atoms @xcite .
the multi@xmath3center molecular integrals over istos can be evaluated by expansion of slater - type orbitals through complete orthonormal basis functions to a new origin @xcite ( see also references therein ) , @xmath15 or by expressing them as a finite linear combination of @xmath16 functions through fourier transform @xcite .
however , infinite series representation formulas arising in expansion method require increasing upper limit of summation as much as possible to converge to exact values with sufficient decimals ( the choice adopted as threshold for the total energy in nonrelativistic variational energy calculation is of order e@xmath303 atomic units , therefore , constitute matrix elements should be accurate to e@xmath310 atomic units ) and presence of spherical bessel functions brings computational difficulties in fourier transform method since they provoke an oscillation @xcite .
( origin ) at ( 0,0,0 ) ; ( 0,0,0 ) ( 7,0,0 ) node[anchor = north east]@xmath17 ; ( 0,0,0 ) ( 0,7,0 ) node[anchor = north west]@xmath18 ; ( 0,0,0 ) ( 0,0,7 ) node[anchor = south]@xmath19 ; \(a ) at ( 2,5,2 ) ; ( b ) at ( 7,4,2 ) ; ( c ) at ( 2.5,2.7,2.5 ) ; ( e ) at ( 4.5,7,2 ) ; ( origin ) circle node [ left ] o ; ( a ) circle ( 3pt ) node [ below left=-0.3 cm and 0.00 of a ] a ; ( b ) circle ( 3pt ) node [ right ] b ; ( c ) circle ( 3pt ) node [ below=0.1 cm and 0.00 of c ] c ; ( e ) circle ( 2pt ) node [ right ] @xmath20 ; ( 2,5,2 ) ( 7,4,2 ) ; ( 2,5,2 ) ( 2.5,2.7,2.5 ) ; ( 7,4,2 ) ( 2.5,2.7,2.5 ) ; ( 0,0,0 )
( 2,5,2 ) node [ black , pos=0.6 , left ] @xmath21 ; ( 0,0,0 )
( 7,4,2)node [ black , pos=0.6 , below ] @xmath22 ; ( 0,0,0 ) ( 2.5,2.7,2.5 ) node [ black , pos=0.75 , below ] @xmath23 ; ( 0,0,0 )
( 4.5,7,2 ) node [ black , pos=0.55 , left ] ; ( 2,5,2 ) ( 4.5,7,2 ) node [ black , pos=0.6 , left ] @xmath24 ; ( 7,4,2 ) ( 4.5,7,2 ) node [ black , pos=0.6 , below ] @xmath25 ; ( 2.5,2.7,2.5 )
( 4.5,7,2 ) node [ black , pos=0.7 , right ] @xmath26 ; ( origina ) at ( 2,5,2 ) ; ( originb ) at ( 7,4,2 ) ; ( originc ) at ( 2.5,2.7,2.5 ) ; the problem of multi@xmath3center integrals evaluation by the use of nstos even much more through insurmountable .
the slater type orbitals with noninteger principal quantum numbers do not have infinite series representation formulas ; they can not be expanded via complete orthonormal basis functions since power series for a function such as @xmath27 , @xmath28 and @xmath29 are not analytic at the origin @xcite , where the symbols @xmath30 , @xmath31 , @xmath32 used to denote the sets of complex , real and natural numbers , respectively . it should be noted that , this also eliminates possibility of applying binomial expansion theorem in order to evaluate the two - center integrals , those are analytically closed form relations may obtain .
therefore , in mathematical point of view evaluation of multi@xmath3center molecular integrals using noninteger principal quantum numbers in slater - type orbitals is an open question .
it is far more than better representation of electronic wave@xmath3function in nonrelativistic electronic structure calculations it is also directly related with solution of the dirac equation in algebraic approximation .
the basis functions to be used in solution of matrix form of the dirac equation are obtained analogously to l - spinors @xcite which are related to the dirac hydrogenic solutions .
their explicit form include power functions @xmath33 , @xmath34 with , @xmath18 is nuclear charge , @xmath35 is speed of light , @xmath36 , respectively .
they can only be represent by finite summation of slater - type orbitals with noninteger principal quantum numbers . in particular , the three@xmath3center integrals are the first set of multi@xmath3center integrals do not have analytically closed form relations .
they have a fundamental importance in the study of molecular systems through @xmath37initio and density functional theory .
they are central to the understanding of multi - center integrals .
they have been commonly studied with methods presented above .
they can also be evaluated through neumann @xcite and laplace expansion @xcite of coulomb interaction in prolate spheroidal coordinates . in a new approach
the two@xmath3center integrals have been calculated by the authors for arbitrary values of parameters and quantum numbers via numerical integration techniques @xcite .
the new relativistic molecular auxiliary functions in prolate spheroidal coordinates are presented @xcite .
they are used to obtain compact form relations for two - electron integrals .
afterwards this idea adapted to calculate overlap integrals via fourier transform formulas @xcite .
the same accuracy , 36-digits , is achieved in both methods .
these are so far only known precise calculations for molecular integrals over nstos .
hence , they are used in this paper to produce benchmark values for three - center one - electron molecular dirac integrals as a first time in the literature .
the _ mathematica _ programming language @xcite is utilized for both analytical and numerical calculations .
taking into account fig . [
fig : geometryfig ] , where depiction of coordinates are given for one electron in a triangular conformation , the three - center nuclear attraction integrals are defined as follows , @xmath38 with , @xmath39 are three arbitrary points of the euclidian space , @xmath40 , @xmath41 .
the neumann expansion for @xmath42 in prolate spheroidal coordinates ( @xmath43 ) , where @xmath44 , @xmath45 , @xmath46 @xcite , @xmath47 here , @xmath48 , @xmath49 are first and second kind associated legendre functions @xcite , @xmath50 refers to lesser and greater of @xmath51 , respectively , is utilized in order to obtain expressions for the three - center nuclear attraction integrals over slater - type orbitals , where the principal quantum numbers are free from specifications ; @xmath52 here , @xmath53^{1/2}}\left ( \frac{r}{2}\right)^{n+n'+1}\end{gathered}\ ] ] are the normalization constants and @xmath54 coefficients @xcite @xmath55 are the integration over azimuthal angle .
the symbol @xmath56 may have the value @xmath57 and is determined by the product of the signs @xmath58 and @xmath59 ( the sign of zero is regarded as positive ) .
the symbols @xmath60 may have the values @xmath57 and 0 : if among the indices @xmath58 , @xmath59 and @xmath61 there occurs a value equal to zero , then @xmath60 is also zero ; if all the indices differ from zero , @xmath62 and the sign is determined by product of the signs @xmath58 , @xmath59 , @xmath61 .
thus , the coefficients @xmath54 differ from zero only with the values @xmath63 , @xmath64 .
+ since it is assumed axes of prolate spheroidal coordinate system centered on a , b substitutions in eqs .
( [ eq : threecenteridef ] , [ eq : neumannex ] ) can be written as follows @xcite , @xmath65 @xmath66 @xmath67 the @xmath68 integrals are the auxiliary functions and they are defined as , @xmath69-\nu\left[\frac{1}{2}\left(\zeta-\zeta ' \right)r_{ab } \right]}\\ \times\overline{\mathcal{p}}_{lm } \left ( \frac{1+\xi\nu}{\xi+\nu}\right)\overline{\mathcal{p}}_{l'm ' } \left ( \frac{1-\xi\nu}{\xi-\nu}\right)\\ \times p_{l}^{\vert m \vert}\left(\xi \right)\overline{\mathcal{p}}_{lm}\left(\nu \right)d\xi d\nu\end{gathered}\ ] ] @xmath70-\nu\left[\frac{1}{2}\left(\zeta-\zeta ' \right)r_{ab } \right]}\\ \times\overline{\mathcal{p}}_{lm } \left ( \frac{1+\xi\nu}{\xi+\nu}\right)\overline{\mathcal{p}}_{l'm ' } \left ( \frac{1-\xi\nu}{\xi-\nu}\right)\\ \times q_{l}^{\vert m \vert}\left(\xi \right)\overline{\mathcal{p}}_{lm}\left(\nu \right)d\xi d\nu,\end{gathered}\ ] ] where , @xmath71 are the normalized associated legendre functions . the auxiliary functions @xmath68 can be defined in a simpler form through relation given for product of two normalized associated legendre functions centered on points @xmath72 , @xmath16 in prolate spheroidal coordinates @xcite , @xmath73^{\lambda-\frac{\lambda+\lambda'}{2}}\overline{\mathcal{p}}_{l\lambda } ( \cos \theta _ { a } ) \overline{\mathcal{p}}_{l'\lambda'}(\cos \theta _ { b})\\ = \sum _ { \alpha = -(2\lambda-\lambda)}^{l}\sum_{\beta = \lambda'}^{l ' } \sum _ { q=0}^{\alpha + \beta-2\lambda-\lambda-\lambda'}{g_{\alpha \beta}^{q}(l\lambda , l'\lambda';\lambda)}\\ \times{\left[\frac{({\xi \nu } ) ^{q } } { ( \xi + \nu)^{\alpha } ( \xi -\nu ) ^{\beta } } \right]}.\end{gathered}\ ] ] which is obtained in explicit form by taking advantage of binomial expansion theorem , @xmath74 with , @xmath75 the coefficients @xmath76 occurring in eq .
( [ eq : ptwoanlegendre ] ) are determined by , @xmath77 @xmath78 @xmath79^{1/2}\\ \times f_{(l-\beta)/2}(l)f_{\beta-\lambda}(l+\beta),\end{gathered}\ ] ] where , @xmath80 , @xmath81 and the quantities @xmath82 are the generalized binomial coefficients .
they are given as , @xmath83 with , @xmath84\leq s ' \leq min(s , n)$ ] and @xmath85 are binomial coefficients indexed by @xmath86 and @xmath87 is usually written @xmath88 , respectively . the eqs .
( [ eq : jgenauxiliary ] , [ eq : kgenauxiliary ] ) are , therefore , obtained as follows , @xmath89=\sum_{\alpha\beta q}g_{\alpha \beta}^{q}(l\lambda , l'\lambda';\lambda)\\ \times \left[\begin{array}{cc } \mathcal{j}^{lm , q}_{n-\alpha , n'-\beta}\left(\zeta,\zeta',r_{ab},\xi_{c } \right ) \\
\mathcal{k}^{lm , q}_{n-\alpha , n'-\beta}\left(\zeta,\zeta',r_{ab},\xi_{c } \right ) \end{array } \right],\end{gathered}\ ] ] with , @xmath90\\ = \int_{{\tiny \left[\begin{array}{cc } 1 \\ \xi_{c } \end{array } \right]}}^{{\tiny \left[\begin{array}{cc } \xi_{c } \\ \infty \end{array } \right]}}\int_{-1}^{1}{\left(\xi\nu \right)^{q}\left(\xi+\nu \right)^{n-\alpha}\left(\xi-\nu \right)^{n'-\beta}}\\ \times e^{-\xi\left[\frac{1}{2}\left(\zeta+\zeta ' \right)r_{ab } \right]-\nu\left[\frac{1}{2}\left(\zeta-\zeta ' \right)r_{ab } \right]}\\ \times \begin{bmatrix } \mathcal{p}_{l}^{\vert \lambda \vert } \left(\xi\right ) \\
\mathcal{q}_{l}^{\vert \lambda \vert } \left(\xi\right ) ] \end{bmatrix}\overline{\mathcal{p}}_{l \vert \lambda \vert}\left(\nu \right)d\xi d\nu.\end{gathered}\ ] ] there are no known convergent series representation formulas , free from specifications on parameters for power functions such as @xmath91 , @xmath92 , yet and that poses an obstacle to analytically reduce the @xmath93 , @xmath94 auxiliary functions to one variable @xmath95 auxiliary functions introduced in @xcite
. thus , the solution should be obtained on the basis of numerical methods .
note that , taking advantage of binomial expansion method for terms containing the angular part of slater - type orbitals in order to simplify the expressions increases the number of integrals should be numerically calculated . in _
programming language instead of using eq .
( [ eq : jkgenauxiliary ] ) , direct computation of eqs . (
[ eq : jgenauxiliary ] , [ eq : kgenauxiliary ] ) are faster .
the given relations for auxiliary functions in eq .
( [ eq : jkredauxiliary ] ) are calculated by using different expressions of legendre polynomials and compared according to computational time in fig .
( [ fig : cput ] ) as sample .
the discussions on results are made in the next section .
the literature currently , lack of benchmark values for multi@xmath3center integrals when slater - type orbitals are used .
recently , a robust numerical global@xmath3adaptive strategy with gauss@xmath3kronrod extension has been applied for two@xmath3center integrals through prolate spheroidal coordinates and fourier transform method in @xcite .
benchmark results have been presented for them .
the hermitian properties are thus , represented correctly free from specification on quantum numbers , orbital parameters and internuclear distances . in this study
it is extended for solution of three@xmath3center integrals .
the algorithm described in @xcite has been incorporated into a computer program written in the _ mathematica _ programming language with the included numerical computation packages for solving eqs .
( [ eq : methreecenter ] , [ eq : jgenauxiliary ] , [ eq : kgenauxiliary ] ) .
the _ mathematica _ programming language can handle approximate real numbers with any number of digits and it is suitable for benchmark evaluation .
it is also provides a uniquely integrated and automated environment for parallel computing .
it is allows us to compute the formulas including summations using all cores of pc effectively via * parallelsum * command instead of * sum*. note that , in this study all results are given in atomic units ( a.u . ) . | the authors in their previous papers obtained compact , arbitrarily accurate expressions for two - center one- and two - electron relativistic molecular integrals expressed over slater - type orbitals . in this present study ,
the accuracy limits of given expressions is examined for three - center nuclear attraction integrals , which are the first integral set do not have analytically closed form relations .
they are expressed through new molecular auxiliary functions obtained via neumann expansion of coulomb interaction .
the numerical global adaptive method is used to evaluate these integrals for arbitrarily values of orbital parameters , quantum numbers .
several methods , such as laplace expansion of coulomb interaction , single - center expansion , fourier transformation method , have been performed in order to evaluate these integrals considering the values of principal quantum numbers in the set of positive integer numbers .
this is the first attempts to study the three - center integrals without any restrictions on quantum numbers and in all ranges of orbital parameters .
keywords : : pacs numbers : : ... . |
elastic nucleon - nucleus ( na ) scattering is known now as an excellent means of testing nuclear structure @xcite .
results of the nonrelativistic theory for such scattering confidently can be used as predictive for information required in applied nuclear technology , where large amounts of nuclear reaction data including fission cross sections at intermediate energies are required for various challenging applications such as accelerator transmutation of waste , particularly the elimination of long - lived radioactive wastes with a spallation source , accelerator - based conversion to destroy weapon grade plutonium , accelerator - driven energy production to derive fission energy from thorium with concurrent destruction of the long - lived waste and without the production of nuclear weapon material , and accelerator production of tritium @xcite .
there is also a great need for such information to be the base in analyzes of patient radiation therapy and protection . with basic science
there is the intellectual challenge to go beyond the physics of a single hadron and understand essential aspects of nuclear physics from first principles such as qcd @xcite .
it is generally agreed that the qcd lagrangian involves nonlinear dynamics .
such makes it very difficult then to understand nuclear physics fully from first principles , and so most of nuclear physics phenomena are interpreted in terms of appropriate effective degrees of freedom .
one such view of natural phenomena in terms of energy scales ( @xmath1 ) , divides the nuclear - hadronic scale into the nuclear structure region @xmath2mev and the nucleon and nucleon - nucleon ( nn ) region , with structure and substructure scales @xmath3gev .
this large separation between the hadronic energy scale and the nuclear binding scale poses stringent difficulties to apply nonlinear qcd directly to understand the physics of nuclei .
however , quantitative calculations based on effective quantum field theory ( eqf ) techniques that arise from chiral symmetry provide an alternative approach . at present , this method is being extended to address few- and many - nucleon interactions . when combined with first principle calculations of the low energy constants from qcd , these eqf may provide a consistent qualitative understanding of properties of nuclei and of low to medium energy nuclear scattering .
besides understanding the structure of nuclei from a qcd point of view , it is of interest to understand the dynamical behavior of nucleons in the presence of nuclear matter
. the relevance of modified nn scattering amplitudes in the form of @xmath0-matrices in the nuclear medium with mean fields and pauli blocking is well known .
these nn amplitudes and alternative reductions in the form of @xmath4-matrices have been used with qualitative success in the specification of nuclear densities in stable nuclei and the description of na scattering for projectile energies ( @xmath5 ) below 1gev . for @xmath5 above this limit
we expect significant dynamical changes due to dibaryonic fusion with subsequent fission in the short range region of nn subsystems @xcite . hitherto quite independently , several groups successfully described intermediate energy na scattering using two nucleon @xmath4- or @xmath0-matrices as driving effective interactions @xcite . among them
we distinguish two main philosophies under which a description of na collisions is made and which we specify as the nonrelativistic schrdinger approach and as the relativistic dirac approach @xcite .
common to both is the explicit use , and accurate treatment , of an interacting nn pair in the realm of other nucleons in a nucleus , and the need for an effective nn scattering amplitude known and defined on- and off - shell . for the more recent calculations of nonrelativistic na optical potentials in momentum space @xcite , a consistent treatment of the fully off - shell nn @xmath4- or @xmath0-matrices
have been used . as nearly all available nn potentials
@xcite have been fitted to nn scattering data up to 350mev , a nonrelativistic treatment of the full - folding optical model suffices most na scattering applications consistent with the input nn model .
recently , however , inversion potentials developed to fit elastic nn phase - shifts for @xmath6gev have been used within the full - folding approach for projectile energies as high as 500 mev @xcite . despite the fact that these were nonrelativistic applications , they allowed a better description of data relative to what was found previously using traditional nn potential models
the lack of relativistic kinematics as well as pion production and @xmath7 excitations within the nn pair did not manifest itself as a dramatic limitation of the model , even though the theoretical confidence level was reached .
it is the primary aim of this work to remove these shortcomings and formulate a momentum space full - folding model using _ minimum - relativity_. with the appropriate modifications to the existing analysis programs we expect to extend the confidence level into the gev region and to obtain high quality full - folding optical model results at energies as high as 1.5gev .
extensive studies of na scattering in the context of relativistic dirac models have been made @xcite . while these models are closer to a fundamental formulation with the inclusion of relativistic kinematics and dynamics within a single framework , to date they all have used fitted nn interactions that do not describe the phase shift data well .
this is a serious drawback as it is well established that any folding model requires a high quality nn interaction as input .
the description of na scattering requires only modest nuclear matter densities and thus the need of nn @xmath0-matrices is less demanding .
furthermore only for projectile energies below @xmath8mev are nuclear medium effects predominant with specific treatment of pauli blocking and self consistent mean field effects being crucial . for medium and higher nn energies ,
meson production and intrinsic hadron excitations within the interacting pair are important .
consequently the @xmath4-matrix no longer is unitary in the elastic channel .
however as the low energy @xmath4- and @xmath0-matrices are well defined within potential scattering theory , we seek a continuation of nn potential models with an nn optical model potential ( nnomp ) .
we have devised and generated nnomp for @xmath9 gev and applied them to na elastic scattering .
the calculations are based upon a relativistically corrected full - folding optical model in momentum space that is an extension of a nonrelativistic predecessor @xcite .
since the lorentz contraction scales as the ratio of the projectile energy to its mass , it requires nucleon projectile energies above @xmath10mev to have sizeable contraction effects .
it is safe to include relativistic kinematics in folding calculations when the projectile energy surpasses 300mev .
relativistic kinematics is widely used in pion - nucleus scattering @xcite .
an excellent discussion on some aspects about relativistic kinematics in na scattering is given in the review by ray , hoffmann and coker @xcite .
one of the important advantages of finding a nonrelativistic na optical model potential is its well defined structure in terms of interacting nucleon pairs .
this framework has been remarkably successful in the study of low energy scattering with its link to the nuclear shell model of single particle bound states .
the combination of target correlations and high quality nn interactions has provided a good first order description of the na dynamics .
this nonrelativistic theoretical approach has been applied with success for many energies and targets and now it is timely to incorporate relativistic corrections so that calculations can be made with energies as high as 1.5gev .
a brief summary of the current full - folding optical model is given in section ii and in which the points where relativistic kinematic corrections are to be made are stressed .
the current experimental situation of nn phase shift analysis up to 3gev is discussed in section iii .
we describe also how the nnomp are determined by the nn phase shift data . in section
iv we set the framework of the @xmath0-matrix calculations and specify minimal relativity considerations . in section v some details of the full - folding calculations and various applications are presented .
we discuss the role of the fermi motion in the nn effective interaction and analyze the sensitivity of na scattering observables upon the use of alternative approaches for the relativistic kinematics .
we also examine total cross sections for nucleon elastic scattering and differential observables at beam energies from the hundreds of mev up into the gev regime .
section vi contains a summary and the main conclusions of our study .
finally , we have included two appendices . in appendix
a the relativistic kinematics transformations in the context of the full - folding approach are outlined as extracted from the articles by aaron , amado and young @xcite and by giebink @xcite . in appendix
b we present the algorithm we have used to determine the nnomp from data .
in the nonrelativistic theory of the optical model potential , the coupling between the projectile and the target nucleus in the elastic channel is given by the convolution between a two - body effective interaction and the target ground - state mixed density .
in the projectile - nucleus c.m .
frame , the collision of a projectile of kinetic energy @xmath11 is described by the optical potential @xmath12 which in a momentum space representation is expressed as @xcite @xmath13 where @xmath14 represents target ground state single - particle wave functions of energy @xmath15 , and @xmath16 restricts the sum to all levels up to the fermi surface @xmath17 .
the two - body @xmath18 matrix is evaluated at starting energies @xmath19 consisting of the mass of the projectile @xmath20 , its kinetic energy @xmath11 , the mass of the target nucleon @xmath21 and its binding energy @xmath15 .
the subscript a+1 indicates matrix elements in the projectile - nucleus c.m .
frame and recoil effects have been neglected for simplicity . in the most recent full - folding calculations of the optical potential for na scattering , where explicit medium effects are incorporated in the two - body effective interaction , the @xmath18-matrix
is represented by an infinite nuclear matter @xmath0-matrix @xcite . in the absence of these medium modifications ,
the @xmath18-matrix has usually been approximated by the two - body scattering matrix associated with the collisions of two free particles @xcite .
none of these full - folding approaches incorporate the necessary relativistic kinematics needed for high energy processes .
the above expression for the optical potential requires the @xmath18 matrix in the projectile - nucleus c.m .
however , most practical two - body potential models are designed to account for the scattering data in the two - body c.m . , where a one - body wave equation ( a schrdinger kind equation ) is used to construct the realistic two - body bare potential fit to elastic scattering and ground state data .
the practical problem that emerges then is how to make use of this description to extract the needed effective interaction in the projectile - nucleus c.m . with an adequate account of relativistic effects
this has been a long standing problem in nuclear research and various approaches have been proposed and discussed elsewhere @xcite . in the procedure followed here we retain the dynamical structure of the optical potential as expressed by eq .
( [ uff ] ) and identify , _ via _ a lorentz boost , the corresponding kinematical variables involved in the two - body collision .
thus , the transformation of the @xmath18 matrix from the two - body ( 2b ) to the projectile - nucleus ( a+1 ) c.m .
frame can be done by considering three separate aspects .
first , as lorentz invariance of the flux is required , an overall normalization factor usually referred as the mller flux factor mediates between the scattering amplitudes in the two frames .
second , the kinematics in the projectile - nucleus c.m .
frame needs to be transformed to the two - body c.m .
system , which is the reference frame where the bare potential model is defined . and third , the transformation of the scattering matrix from the two - body to the projectile - nucleus c.m .
frames involves the rotation of the spins , an effect referred as the wigner rotation .
this contribution has been studied in the context of the relativistic ` no - pair ' potential for nucleon - nucleus scattering by tjon and wallace @xcite and was observed to yield rather moderate effects at nucleon energies between 200 and 500 mev .
although there is no statement about the importance of wigner rotation contribution at the higher energies considered here , we shall neglect them in the present work .
consistent with the notation introduced in fig .
[ tmatrix ] for the coupling between the projectile and the target struck nucleon , we denote their respective incoming and outgoing four - momenta @xmath22 when translational invariance is assumed , such is the case of free or particles in infinite nuclear matter , the two - body interaction is characterized by the starting energy @xmath23 and the total momentum of the colliding particles @xmath24 , @xmath25 here the dirac delta - function makes explicit the total three - momentum conservation of the two - body collision , @xmath26 , where @xmath27 and @xmath28 represents the energy in two - body c.m .
, @xmath29 the overall coefficient @xmath30 is the mller flux factor , @xmath31^{1/2 } \;\ ; , \ ] ] where the energies @xmath32 and @xmath33 are on - mass - shell , i.e. @xmath34 and @xmath35 .
what remains to be specified is a lorentz transformation for the relative momenta @xmath36 and @xmath37 . to this purpose
we followed the approaches introduced by aaron , amado and young ( aay ) @xcite and giebink @xcite .
although , both assume a representation of the scattering matrix in the two - body c.m . , they differ in the way the lorentz boost is devised .
details are given in appendix a where we show that in both cases the relative momenta can be cast into the form @xmath38 with @xmath39 and @xmath40 being functions of the momenta of the colliding particles ( eqs .
( [ k_gie ] ) or ( [ k_aay ] ) ) . at low beam energies
both prescriptions meet the nonrelativistic limit @xmath41 , as expected .
of the whole spectrum , low and medium energy nn scattering traditionally is described in terms of hermitian potentials . at medium energies , production processes and inelasticities become possible and several elementary systems composed of nucleons and mesons contribute to nn scattering . at present
there is no high quality description of nn scattering above the inelastic threshold either in terms of qcd or in terms of nucleons and mesons
. a high quality fit of on - shell @xmath4-matrices by means of a potential model is very desirable as it provides extensions of the effective interaction into the off - shell domain and into a nuclear medium , which are important dynamical features in few and many body calculations .
many examples using microscopic optical model potentials for elastic nucleon - nucleus scattering and bremsstrahlung reactions , have shown that it is crucial to have on - shell @xmath4-matrices in the best possible agreement with nn data at all energies .
concomitantly one needs high precision nn data against which one can specify nn interactions . to this purpose
we have relied on a large body of experimental nn data whose parameterization in terms of amplitudes and phase shifts are smooth for energies to 3 gev @xcite .
this is a supposition for the construction of an nn potential above 300 mev .
there are many studies of few and many body problems in the low energy regime @xmath42mev and the results have consequences for any model extension above threshold .
we note in this context that significant off - shell differences in @xmath4-matrices are known to exist among the theoretically well motivated boson exchange models of nn scattering .
it remains difficult to attribute with certainty any particular dynamical or kinematical feature with those differences .
non - locality , explicit energy dependence and features associated with relativistic kinematics are some possibilities .
in contrast , there is the quantum inverse scattering approach by which on - shell @xmath4-matrices can be continued into the off - shell domain .
a specific method is the gelfand - levitan - marchenko inversion algorithm for sturm - liouville equations .
this approach to specify @xmath4-matrices off - shell is appropriate when the physical s - matrix is unitary and the equation of motion is of the sturm - liouville type .
such is valid without modification for nn @xmath4-matrices in the energy regime below 300 mev , and for the unitary part of the s - matrix above that energy . in the spirit of general inverse problems ,
we have extended the available low energy potential by additional complex potential terms which are determined from a perfect reproduction of the experimental data ( here the partial wave phase shift analysis ) for all energies above 300 mev . by
that means nn optical models were generated , separately for each partial wave .
the algorithm we have developed allows studies of complex local and/or separable potentials in combination with any background reference potential @xcite .
here we limit the reference potential to the well known real coordinate space potentials from paris @xcite , nijmegen @xcite ( reid93 , nijmegen - i , nijmegen - ii ) , argonne @xcite ( av18 ) , and from inversion @xcite . to them
we add channel - dependent complex separable potentials with energy dependent strengths @xcite . for a given input data set
there is a unique nnomp within a given potential class .
some of these issues are outlined in appendix b. nn scattering is a long standing problem which has been reviewed often as the database developed .
the low energy data has been analyzed by the vpi / gwu group @xcite for @xmath43 mev , the nijmegen group with the nn phase shift results pwa93 for @xmath44 mev , and by machleidt @xcite giving the bonn - cd-2000 . of these
, the vpi / gwu group has given many solutions over the years , the latest are for energies to 3 gev @xcite .
we have used the solutions sp00 , fa00 and wi00 in our calculations and found results that differ but marginally . thus hereafter in the main we refer solely to the results of calculations based upon sp00 . as with the nn phase shift analysis ,
one boson exchange potentials have received several critical reviews @xcite , including observations that there are small variations between phase shift analysis and potential models below the subthreshold domain @xmath45 mev .
a theoretically stable result would require many quantities , that need be specified _ a priori _ , to be determined by independent sources .
at present that does not seem feasible and all current potentials rely upon fits of many of their parameters to the same data .
all such fits , however , have been made independently of each other and are based upon differing theoretical specifications of the boson exchange model dynamics . above 300 mev , reaction channels open and the elastic channel s - matrix no longer is unitary .
only the @xmath7(1232 ) resonance has a low energy threshold and a relative small width of 120 mev .
therefore it is the only resonance we expect to be obviously visible in the energy variation of the elastic scattering phase shifts . in particular
one notices typical variations in the @xmath46 , @xmath47 , and @xmath48 channels .
otherwise the phase shifts to 3 gev vary smoothly as functions of energy .
together with the strong spin - isospin coupling , this property infers optical potentials that are channel dependent in contrast to the na case for which assumed central and spin - orbit potentials are partial wave independent . the plethora of reaction channels that open to 3 gev , and
the requirement of an nn optical potential prescription , mean that it is an interesting task for a microscopic model to link qcd substructures to nn scattering phase shift functions in analogy to that successful prescription by which na optical potentials have been determined by folding effective interactions . to describe this developing system for @xmath9 gev we used feshbach theory to specify the optical potential .
an important feature of that theory is the projection operator formalism with @xmath49 and @xmath1 subspaces , which divide the complete hilbert space @xmath50 into the elastic scattering channel , the @xmath49 space , and the inelastic and reaction channels , the @xmath1 space .
this infers a complex and separable component in the optical potential with an energy dependent strength .
if a very large number of intermediate states contribute , the effect equates to a local potential operator . a covariant description of nn scattering formally is given by the bethe - salpeter equation @xmath51 where @xmath52 are invariant amplitudes that are based upon all connected two particle irreducible diagrams .
this equation serves generally as an ansatz for approximations . of those ,
the three dimensional blankenbecler - sugar reduction is popular and sufficient for our purpose to define an nn potential @xcite .
the amplitudes are now expressed with the reduced terms and they satisfy a three - dimensional equation @xmath53 taking matrix elements with only positive energy spinors , an equation with minimum relativity results for the nn @xmath4-matrix , namely @xmath54 using the substitutions @xmath55 and @xmath56 a simplified form of the @xmath4-matrix is obtained .
it is the familiar lippmann - schwinger equation @xmath57 the strategic importance of this result is that it defines a sensible continuation of a @xmath4-matrix , constrained on - shell by the experimental data and phase shift analysis , into the off - shell domain as required by the full - folding optical model .
thus we do not rely primarily on a fundamental theoretical result but rather on experimental nn data and the moderate sensitivity of na scattering to alternative off - shell continuations .
it is an important result also of this analysis that on - shell equivalent nn optical model potentials yield very similar na scattering observables irrespectively of differences in the off - shell domain and the constraint off - shell continuation defined with eqs.([eqn_iii.8]-[eqn_iii.12 ] ) .
a crucial step in the description of na scattering processes has been the definition of an effective interaction based on the bare two - body interaction in free space .
such has been the philosophy of the early ( local ) folding models and the most recent nonrelativistic full - folding models @xcite . the effective interaction , @xmath58 in eq . ( [ uff ] ) ,
is obtained in the framework of brueckner - bethe - goldstone for the @xmath0-matrix .
the extension of this approach to high energy applications requires a minimal account of relativistic corrections . along this line
we have followed the discussion by brockmann and machleidt @xcite , where a relativistic three - dimensional reduction of the bethe - salpeter equation is used to describe the interaction between nucleons in the nuclear medium .
if only matrix elements between positive - energy spinors are taken , then the medium - modified invariant amplitude in an arbitrary frame reads ( cf .
( a17 , a18 ) in ref .
@xcite ) @xmath59 here the momentum @xmath24 represents the momentum of the pair with respect to the background , and @xmath60 the pauli blocking operator which projects onto unoccupied intermediate states . for the above expression angle averages have been used , i.e. @xmath61 , and the @xmath62 invariant has been defined as @xmath63 .
this approach , in the context of the dirac - brueckner - hartree - fock approximation , has been applied with reasonable success to the study of infinite nuclear matter @xcite as well as finite nuclei ground state properties @xcite .
an appealing feature of the above equation for @xmath64 is its direct connection with the bare nn potential model in free space . indeed , adopting the same definitions as in eqs .
( [ eqn_iii.10 ] ) and ( [ eqn_iii.11 ] ) , @xmath65 and @xmath66 the following equation for the @xmath0-matrix is obtained @xmath67 the above dirac - brueckner approach differs in a non - trivial way from the conventional nonrelativistic brueckner approach .
the density dependence of the one - boson - exchange interaction by means of effective dirac spinors and the explicit relativistic kinematics are features which have no counterpart in the traditional brueckner approach .
however , with the suppression of these relativistic dynamical effects we can extract a minimum of relativistic features needed in the nonrelativistic model . in our approach selfconsistency is demanded with the following choice for the quasi - particle spectrum , @xmath68 as in the usual brueckner - bethe - goldstone approach , the quasi - potential @xmath69 is obtained selfconsistently with the use of the continuous choice at the fermi surface . as a first check , we consider the case of two nucleons interacting in free space , for which the pauli blocking operator becomes the identity and the nuclear selfconsistent field vanishes .
furthermore , if the interaction is described in the pair c.m .
( @xmath70 ) then the @xmath0-matrix corresponds to the free scattering matrix @xmath71 as described by the bbs equation ( cf .
. [ eqn_iii.12 ] ) .
this limit is immediately verified upon substitution of @xmath28 by @xmath72 , with @xmath73 the on - shell c.m . relative momentum .
thus , @xmath74 which is the nonrelativistic lippmann - schwinger equation with the pole at the relativistically correct momentum .
the other case of interest is infinite nuclear matter , where the relativistic structure of the quasi - particle spectrum introduced in eq .
( [ spectrum ] ) can be assessed considering the following power expansion in terms of @xmath75 @xmath76)\;.\ ] ] if we substitute @xmath77 by the angle averaged quantity @xmath78 and proceed similarly with @xmath62 , the similarity of the energy denominator in eq .
( [ gmat ] ) with the one obtained using the nonrelativistic propagator is evident .
an estimate of the accuracy of the above approximation at normal densities can be made considering @xmath79 1/10 .
in such a case the above form of @xmath80 yields a propagator equivalent to its nonrelativistic counterpart with an accuracy better than 1% .
this result supports the use of the nonrelativistic selfconsistent scheme to obtain the quasi - potentials . with the above considerations , the calculation of the two - body effective interaction needed in the full - folding optical potential proceeds with the proper choice of the @xmath62-invariant in eq .
( [ gmat ] ) , and consistent with the starting energy @xmath81 defined in eq .
( [ uff ] ) .
since @xmath81 represents the total pair energy of interacting nucleons with total momentum @xmath24 , the @xmath62-invariant is simply given by @xmath82 .
a few remarks about the actual implementation of the computational procedures , aimed to obtain the full - folding optical potential @xmath12 , are worthy to note .
first of all , we use the slater approximation for the nuclear mixed density @xmath83 in eq .
( [ uff ] ) .
this approximation has been discussed in the past @xcite and becomes particularly suitable for the present applications , where we rely on the same point nuclear densities for the target ground state .
furthermore , we make use of the infinite nuclear matter @xmath0-matrix to represent the effective interaction between the projectile and the target nucleon @xcite . as a result ,
the @xmath0-matrix evaluated at a fermi momentum @xmath84 is folded with the target ground state density @xmath85 at a local momentum @xmath86 . the slater approximation suggests the ansatz @xmath87 , where the local momentum @xmath88 is determined from the density by @xmath89 .
these considerations yield for the full - folding optical model potential @xmath90 where @xmath91 and @xmath92 are the local proton and neutron point densities respectively , and @xmath93 represent the off - shell fermi - averaged amplitudes in the nn channel . for a particular channel this amplitude depends on the nuclear matter density _ via _ @xmath84 ( implicit in @xmath0 ) , and the local momentum @xmath88 which sets the bounds for the fermi motion of the target nucleons .
this is expressed as @xmath94 where @xmath95 and the relative momentum @xmath36 and @xmath37 are obtained following each of the two relativistic prescriptions discussed in appendix a. we stress at this point that these amplitudes are calculated fully off - shell and that no assumption is made regarding the coordinate space structure of the @xmath0-matrix . with these considerations
the full - folding optical potential becomes a genuine non - local operator .
its use in the schrdinger equation involves integro - differential equations which are solved exactly within numerical accuracy .
the nuclear matter calculations for the @xmath0-matrix were done with fully self - consistent fields at various values of @xmath84 .
we have considered the inversion potentials based upon the sp00 phase - shifts solution with nijmegen - i , -ii and reid-93 reference potentials ( sp00-nij1 , sp00-nij2 and sp00-re93 respectively ) , argonne reference potential ( sp00-av18 ) , paris reference potential ( sp00-pari ) and gelfand - levitan - marchenko inversion reference potential ( sp00-invs ) . to disclose some of the features exhibited by the fermi - averaged effective interaction in the context of the full - folding approach
, we analyze the @xmath96 averages in the proton - proton ( @xmath97 ) and neutron - proton ( @xmath98 ) channels . here
we focus on the on - shell forward matrix element . in this case
the @xmath93 amplitude depends on the na projectile momentum @xmath99 , the fermi momentum @xmath84 , and the local momentum @xmath88 . in the case of the free @xmath4-matrix approach for
the interaction we set @xmath100 but allow for the variation of @xmath101 implied by the local nuclear density to account for the fermi motion in the nucleus ( @xmath102 ) . in the case of a @xmath0-matrix element we set @xmath103 . in fig .
[ g_av18 ] we show the real and imaginary components of @xmath104 and @xmath105 , based on the sp00-av18 nnomp , as functions of the beam momentum and for the sequence @xmath106 @xmath107 . in each frame
we draw the case @xmath108 @xmath107 with a thick solid curve ; the following values of @xmath88 depart sequentially from this reference curve .
the upper and lower frames correspond to @xmath0- and @xmath4-matrix results respectively . to assess the role of the imaginary part of the nnomp , the results based on the full model are shown in the four leftmost frames with solid curves while the results where the imaginary part is suppressed are shown in the four rightmost frames with dashed curves .
the two relativistic kinematics prescriptions giebink and aay yield almost indistinguishable results for the amplitudes .
a comparison of results in the upper and lower frames indicates more dispersion due to the fermi motion within the @xmath0-matrix than in the @xmath4-matrix approach . this is a feature
which is more pronounced in the real than the absorptive component of the amplitude .
this is a clear indication of the role of the self - consistent fields in the fermi - averaged quantities . in the context of the full nnomp
however , the manifestation of this sensitivity becomes diminished at projectile energies above 500 mev as the real component of all amplitudes for @xmath99 above @xmath1095 @xmath107 are significantly smaller than their imaginary counterparts .
such is not the case at the lower energies , where both the real and imaginary components of the amplitude become comparable .
thus , we do expect more sensitivity under fermi motion in the context of the @xmath0 matrix at projectile energies below 500 mev .
another feature that emerges from fig .
[ g_av18 ] is the sensitivity of the absorptive part of @xmath96 to the presence of the imaginary part of the nnomp . indeed , when this part is suppressed , the absorptive component of the amplitude saturates above @xmath1095 @xmath107 , in contrast with the full nnomp where the trend of this absorption is to increase .
the manifestation of this feature becomes clear when the fermi average enters fully off - shell in the evaluation of the full - folding potential , as discussed in the following section . a global assessment of the full - folding model to the inclusion of relativistic kinematics and features of the underlying nn potential model over a wide energy range
is made first by studying total cross sections for neutron - nucleus elastic scattering . in fig .
[ xtotalx4 ] we show the measured @xcite and calculated total cross sections for neutron elastic scattering from @xmath110o , @xmath111ca , @xmath112zr and @xmath113pb at beam energies ranging from 100 mev up to 1 gev . these cross sections are obtained by solving the scattering equations using full - folding optical potentials calculated as in ref .
@xcite , modified to include the kinematics discussed in this work . for completeness in this comparison we have considered the full nnomp within the @xmath0- ( solid curves ) and @xmath4-matrix ( dashed curves ) approaches . in order to assess the role of the absorptive contribution of the nnomp , we have also included results suppressing the imaginary part .
these results are shown unmarked , whereas those using the full nnomp have been labeled with a triangle at their right end .
the data are represented with open circles . from this figure
we observe a remarkable agreement between the full nnomp @xmath0-matrix results and the data , particularly at energies above @xmath109200 mev .
such is not the case for the @xmath4-matrix approach , or when the imaginary part of the bare nn potential is suppressed . in the former case
the lack of nuclear medium effects becomes pronounced for projectile energies as high as 500 mev in the case of pb , but less for lighter nuclei .
conversely , the imaginary part of the nnomp is crucial for the adequate description of cross section data at beam energies above 400 mev , as observed when comparing the labeled and unlabeled curves .
total reaction cross sections for proton - nucleus elastic scattering at high energies are of increasing interest , particularly with current trends using spallation facilities with high energy beams .
thus , we calculated total reaction cross - section for elastic proton scattering , at beam energies up to 1.5 gev , and compare them with available data @xcite . in strict analogy with the previous applications , and use of notations as in fig .
[ xtotalx4 ] , in fig .
[ xreactx4 ] we show @xmath0- ( solid curves ) and @xmath4-matrix ( dashed curves ) results wich used the full or suppressed the imaginary part of the nnomp . here
again we have marked with triangular labels those results based on the full nnomp , and left unmarked the ones with the imaginary part suppressed .
the data are represented by open circles .
the results shown in fig .
[ xreactx4 ] used the sp00-av18 nnomp . for this particular observable and energies of our study
, we find very similar results among all the nnomp i.e. irrespectively of the reference potential .
again we find little differences between the @xmath0- and @xmath4-matrix approaches above @xmath109500 mev .
thus , medium effects in the interaction are rather weak ( albeit not neglegible ) at these higher energies .
such is not the case below 400 mev , particularly in the case of pb , where a clear departure of the @xmath4-matrix with respect to the @xmath0-matrix results is observed .
however , these differences are smaller than the ones due to the presence of the imaginary part of the nnomp .
in fact , we can see clear differences within the @xmath0-matrix approach by including and suppressing the imaginary part of the nnomp at energies above 500 mev . since reaction cross section data are scarce above 700 mev ( only two data points for all four targets ) , the curves ( marked with triangles ) constitute a high energy prediction of our work .
another feature observed from figs .
[ xtotalx4 ] and [ xreactx4 ] is the trend of all cross sections to reach a plateau above 650 mev .
we have scrutinized more closely this feature and find that both @xmath114 and @xmath115 exhibit an almost linear dependence with @xmath116 , with target masses 16@xmath117208 . to illustrate this point we show in fig .
[ xa23 ] the calculated @xmath114 and @xmath115 as function of @xmath116 .
the solid curves represent results from @xmath0-matrix full - folding results , based on the sp00-av18 full nnomp , at various energies ; the dotted curves represent the corresponding results without the imaginary part of the nnomp .
the dashed curves serve as reference and correspond to the parametric forms @xmath118 and @xmath119 in barn units .
notice , only the full nnomp results feature a moderate deviation from the parametric forms shown , which is not the case in calculations where the imaginary part of the nnomp was suppressed ( dotted curves ) .
thus , the absorptive part of the nnomp inhibits the energy dependence of the total cross sections above 700 mev .
differential cross sections and spin observables for elastic scattering remain a challenge for any microscopic theory .
an assessment of alternative relativistic kinematics is made with a comparison of these observables using giebink and aay relativistic approaches .
these results are illustrated in fig .
[ xaq_ca1040_kin ] , where the differential cross section @xmath120 , analyzing powers @xmath121 and spin rotation functions @xmath1 for p+@xmath111ca elastic scattering at 1.04 gev are shown as function of momentum transfer @xmath122 .
we have used @xmath0-matrices together with giebink ( solid curves ) and aay ( dashed curves ) kinematics .
this figure shows the case where the differences should be most pronounced .
indeed , all the other cases show nearly complete overlap between the two cases
. nevertheless , fig .
[ xaq_ca1040_kin ] shows that the two prescriptions yield similar results for all observables .
some differences can be seen in @xmath121 , at momentum transfers above 1.5 @xmath107 , where the aay lie slightly above the giebink kinematics results .
we selected for all further calculations the giebink kinematics .
next , we study six nnomp in na elastic scattering at various energies with the full - folding model using the @xmath0-matrix approach .
we have chosen scattering of protons from @xmath111ca and @xmath113pb for which there is a large body of high precision data over a wide energy range @xcite .
the applications used the full nnomp with reference to sp00-nij1 , -nij2 , -pari , -av18 , -re93 and -invs solutions .
as in most cases the differences among these reference nn potentials are moderate , we have chosen all curves with a single pattern .
this also helps to illustrate the level of sensitivity of na scattering upon on - shell equivalent potentials with different off - shell behavior .
considering that the results presented here correspond to parameter free calculations , the overall description of all elastic scattering data is remarkably good .
only a close examination of the results shows the limitations of our approach . in fig .
[ xaq_ca3 - 5 ] for p+@xmath111ca scattering at 300 , 400 and 497.5 mev , for instance , we observe a very good agreement between the full - folding model results for all observables and momentum transfers above @xmath1091 @xmath107 .
however , some discrepancies remain in the description of low momentum transfer data . in fig .
[ xaq_ca6 - 10 ] are shown p+@xmath111ca results for energies between 650 and 1040 mev .
we observe two curves slightly separated from the rest .
such curves correspond to the sp00-av18 and sp00-pari nnomp , affecting mainly the differential cross section and analyzing powers at 800 and 1040 mev . the possible cause of this feature , and the fact that this is more pronounced in the case of ca than in pb is not fully understood but numerical reasons can not be ruled out . a similar trend is observed in fig .
[ xaq_pb3 - 5 ] for p+@xmath113pb scattering .
these limitations have already been noticed in previous nonrelativistic full - folding calculations .
thus , the identification of a unique missing element in the approaches , to account for the low @xmath122 failures remains an open issue .
some possibilties have been discussed elsewhere @xcite . in figs . [
xaq_pb3 - 5 ] and [ xaq_pb6 - 10 ] we show results for p+@xmath113pb scattering at energies between 300 mev and 1 gev . with the exception of the 400 and 497.5 mev cases ,
the cross section is very well described by our calculations . with respect to the spin observables
, there is a tendency to lose structure relative to the measured values at energies above 650 mev .
the calculation of optical potentials , within the full - folding approach used herein , are made without refuge in assumptions either about the local structure of the nn effective interaction or about the final structure of the na coupling . in fact ,
these potentials are treated as non - local operators and are the result of a detailed account of the nn effective interaction off - shell .
in contrast , the optical potentials in the nonrelativistic impulse approximation , discussed in ref .
@xcite , are assumed local and constructed with only on - shell @xmath4-matrix elements as effective interaction .
important differences have been observed between these two approaches when applied to intermediate energy ( 200 - 400 mev ) na scattering .
an illustration of these differences is made in fig .
[ xaq_ff_trho ] for 800 mev p+@xmath113pb elastic scattering , where the differential cross section , analyzing powers and spin rotations are shown as function of the momentum transfer .
the @xmath0-matrix full - folding results are represented with solid curves .
the @xmath4-matrix full - folding results are represented with long - dashed curves , whereas short - dashed and dotted curves are used to represent the results of the off - shell @xmath123 and on - shell @xmath123 results respectively .
clear differences can be seen between the @xmath0- and @xmath4-matrix results , particularly for the spin observables .
these are pronounced for @xmath124 1.5 @xmath107 which illustrates the level of sensitivity to medium effects at these momentum transfers and for this particular case . within the same @xmath4-matrix approach
however , the full - folding results are similar to those obtained within the @xmath123 approximation .
the extent of these sensitivities is comparable to contributions from short range correlations @xcite .
the full - folding calculations presented here are first of their kind to be tested at energies as high as @xmath1091.5 gev .
differences are observed between these and previous results , particularly in the 800 mev applications at forward angles @xcite .
apart from the locality issue in the na coupling and the treatment of the nn effective interaction off - shell , additional effects have been included in those local potential calculations and which may cause variations between the results . particularly relevant
seem to be the short range correlations and electromagnetic spin - orbit contributions to the na coupling .
a careful assessment of these effects in the context of the full - folding approach is needed .
considering that calculations in the @xmath123 scheme are much simpler than those in the full - folding approach , the former becomes quite suitable for exploratory purposes .
quantitative comparisons , however , do require the inclusion of medium effects within the full - folding approach .
the nonrelativistic full - folding optical model approach for na scattering has been extented into the relativistic regime .
kinematical issues involving the off - shell lorentz boost of the colliding particles between the nn and the na c.m .
reference frames have been addressed .
also , explicit nuclear medium effects have been incorporated with the use of microscopic nn effective interaction as obtained in the framework of nuclear matter @xmath0-matrix using an nnomp which fully accounts for the inelasticities and isobar resonances at nucleon energies as high as @xmath1093gev .
the nuclear matter @xmath0-matrices were obtained considering both pauli blocking and selfconsistent nuclear fields as in the traditional brueckner theory .
minimal relativity corrections were extracted from the brockmann and machleidt approach to relativistic nuclear matter .
effects arising from wigner rotations and electromagnetic spin - orbit corrections were not included .
the study considered both @xmath4-matrix and _ in - medium _ self - consistent @xmath0-matrix approaches . with the inclusion of relativistic kinematics corrections , in conjunction with a realistic description of nn resonances and inelasticities by means of an nnomp
, we obtain a good description of both the total and the differential scattering observables for na elastic scattering . the results exhibit a weak sensitivity to the choice of the relativistic approach aay or giebink to correct the kinematics .
we also observe that medium effects are significant over the whole energy range of our study , although they are rather weak above 400 mev .
in contrast , the inelasticities of the nn interaction become important above 400 mev as was observed particularly in the description of the total cross sections .
although our study allows a reasonable description of the differential observables at energies as high as 1 gev in na scattering , specific details remain to be explained ; notably the low @xmath122 behavior of the spin observables in the 400 - 500 mev range . as our primary effort
has been to provide a parameter free nonrelativistic framework for the study of na elastic scattering with a minimal account of relativistic effects at these high energies , a systematic study of various other effects has not been pursued in order to keep our discussion focused . however , future work will scrutinize the systematics of the calculated observables under the use of alternative densities , mixed density representations , electromagnetic effects and higher order correlations .
off - shell effects arising from the non - locality range of the separable description of the nnomp above pion threshold also require further investigation . nevertheless , the level of agreement we have achieved , within the nonrelativistic approach , is comparable to what has been obtained within relativistic approaches .
the study presented is limited in its value as we have not used a covariant two - body and ( a+1)-body dynamics .
the approach adopted is a practical one and is largely motivated and justified by the good agreement of the numerical results with data .
thus we claim , that use of minimal relativity in conjuction with nn optical model interactions , which fully account for the inelasticities and isobar resonances above the pion threshold , yield quantiative descriptions of na scattering up to energies of 1.5 gev .
we have used two prescriptions for the relativistic kinematics involved in the transformation of the two - body colliding momenta between the pair c.m . and the projectile - nucleus c.m .
these schemes have been developed by giebink @xcite and by aaron _
the latter has also been obtained by imposing time reversal invariance on the scattering amplitude @xcite .
following giebink in the context of a manifestly lorentz invariant two - body transition amplitude , let @xmath125 and @xmath126 be the projectile and struck - nucleon four - momenta in the projectile - nucleus c.m .
the corresponding momenta in the exit channel are represented with @xmath127 and @xmath128 . in giebink s
approach the total four - momentum is conserved , @xmath129 which , to be valid , requires each particle to be off mass - shell .
the relative momenta in the two - body c.m .
is readily obtained by applying a lorentz transformation to momenta with a single boost @xmath130 .
this boost is obtained from the invariant @xmath131 and from @xmath132 with this velocity , a direct calculation yields for the incident ( @xmath36 ) and outgoing ( @xmath133 ) relative momenta @xmath134 in these expressions the subscript ` r ' denotes the on mass - shell relative energy @xmath135 where @xmath20 and @xmath21 represent the masses of the projectile and struck - nucleon respectively . in all the above expressions
the magnitude of the relative momentum @xmath136 is needed .
it can be shown that @xmath137 where the @xmath138-function is defined as @xmath139 notice that the @xmath138-function in eq .
( [ k2_gie ] ) is evaluated at the off - mass - shell invariants @xmath140 and @xmath141 .
the actual implementation of giebink s procedure faces the difficulty of improper lorentz transformations which occur when @xmath142 in eq .
( [ s ] ) becomes negative ; this happens for very large momenta @xmath143 .
however , as the bound - state wavefunctions of the struck nucleons confine the momentum distribution of @xmath144 to magnitudes below @xmath1092 @xmath107 , such improper contributions occur for very large @xmath145 and @xmath146 , affecting only far off - shell matrix elements in @xmath147 . a way to circumvent this difficulty is to restrict @xmath142 in eq .
( [ s ] ) near on - mass - shell .
thus , we approximate @xmath148 averaging the fermi motion of the target nucleons ( @xmath144 ) allows the simplification @xmath149 consistent with the above , the following forms are obtained for the relative energies @xmath150 clearly @xmath151 , which is a handy result for the energy denominators in eq .
( [ k_gie ] ) .
the full - folding calculations following giebink relativistic kinematics were made using @xmath152 and @xmath153 , with @xmath154 the nucleon mass . in the relativistic prescription developed by aaron and collaborators
@xcite , the relative momenta for the incoming and outgoing channels require different boost velocities . in each channel
the particles are set on - mass - shell and the corresponding boost is represented by eq .
( [ beta_gie ] ) .
the resulting relative momentum exhibits the same structure as the one given by eq .
( [ k_gie ] ) but with the substitution @xmath155 and @xmath156 .
a direct calculation yields for the magnitude @xmath157 where @xmath158 an analogous result is obtained for the outgoing channel .
consider that there are three distinct hamiltonians @xcite .
they are the _ reference _ hamiltonian @xmath159 , a _ projected _ hamiltonian @xmath160 , and a _ full optical model _
hamiltonian @xmath161 .
the first of these , the reference hamiltonian @xmath162 , invokes a given potential @xmath163 for which one can find schrdinger equation reference solutions .
the physical outgoing solutions @xmath164 of @xmath159 we suppose yields a unitary s - matrix .
we assume further that this hamiltonian is completely specified such that evaluation of any quantity , wave function , s - matrix , k - matrix _ etc .
_ is facilitated .
the feshbach projection operator formalism gives the projected hamiltonian , @xmath165 .
we presuppose completeness , @xmath166 , and a finite rank representation of the q space @xmath167 with the q space basis functions @xmath168 interpreted as doorway states . with these doorway states
we make the link between the qcd and the hadronic sectors ; the latter encompassing nucleons , mesons and other free particles .
thus we assume that meson creation / annihilation occurs only in the highly nonlinear qcd sector so that q space wave functions are projections of such processes onto hadronic particle coordinates .
the third of our hamiltonians , the _ full optical model _
hamiltonian , comprises the reference hamiltonian @xmath159 and the proper optical model potential @xmath169 . that potential is complex and nonlocal , and taken to be separable of finite rank , @xmath170 .
the schrdinger equation specified with @xmath161 has regular physical solutions @xmath171 whose asymptotic boundary conditions we deem to match with the _ experimental _ elastic channel s - matrix .
specifically , for these experimental s - matrices we have used the continuous solutions sp00 from vpi / gwu @xcite . the reference potential @xmath163 and separable potential form factors are still to be specified in detail with the application . to obtain the optical potential on the basis of a given reference potential ,
we express first the solutions of the projected hamiltonian in terms of the reference hamiltonian and the _ a priori _ defined q space projector .
the lippmann
schwinger equation , @xmath172 implies that @xmath173 the solutions of eq.([eqn_iii.19 ] ) can be written in terms of @xmath174 as @xmath175 wherein one can identify a separable potential @xmath176 note then that definition of q space gives a specification of the separable strengths @xmath177 that is unique .
the resultant eq.([eqn_iii.22 ] ) has the form of a first order born approximation but in fact it is an exact result .
to proceed , initially we abandon the exactitude of eq.([eqn_iii.22 ] ) and require the strength matrix , @xmath178 to be constrained asymptotically by the experimental s - matrix of the full hamiltonian schrdinger equation , _
i.e. _ asymptotically we induce @xmath179 .
this implies that complex optical model strengths @xmath180 emerge as a result of matching to riccati - hankel functions and non unitary s - matrices with @xmath181 the strengths @xmath180 then can be simply determined from the linear system of equations @xmath182 to reinforce a lippmann schwinger equation , with the experimental s - matrix as boundary condition or equivalently with strengths @xmath180 from eq.([eqn_iii.26 ] ) , a transformation of the separable potential eq.([eqn_iii.23 ] ) is made .
this is achieved with @xmath183 which contains the separable potentials as defined with eq.([eqn_iii.23 ] ) but whose strengths now are solutions of eq.([eqn_iii.26 ] ) . as the transformation eq.([eqn_iii.27 ] ) contains integration of orthonormal functions , only strengths are altered . using this optical model in the full hamiltonian ,
physical solutions are obtained with reference solutions @xmath174 and greens function @xmath184 of the reference hamiltonian @xmath159 by means of the lippmann
schwinger equation @xmath185 , physics dept . , the george washington university , http://gwdac.phys.gwu.edu/. r. brockmann , and r. machleidt , phys .
c * 42 * , 1965 ( 1990 ) .
h. mther , r. machleidt , and r. brockmann , phys .
c * 42 * , 1981 ( 1990 ) .
r. w. finlay , w. p. abfalterer , g. fink , e. montei , t. adami , p. w. lisowski , g. l. morgan , and r. c. haight , phys .
c * 47 * , 237 ( 1993 ) .
r. f. carlson , atomic data and nucl . data tables * 63 * , 93 - 116 ( 1996 ) .
l. ray , and g. w. hoffmann , phys . rev .
c * 31 * , 538 ( 1985 ) ; d. a. hutcheon _ et al . _ ,
a483 * , 429 ( 1988 ) ; e. bleszynski _ et al .
c * 37 * , 1527 ( 1988 ) ; g. w. hoffmann _
et al . _ ,
47 * , 1436 ( 1981 ) ; a. rahbar _
et al . _ ,
lett . * 47 * , 1436 ( 1981 ) ; r. w. fergerson _ et al .
_ , phys . rev . *
c33 * , 239 ( 1986 ) .
c * 33 * , 239 ( 1986 ) ; g. d. alkhazov _ et al . _ ,
a274 * , 443 ( 1976 ) ; g. d. alkhazov _ et al .
_ , phys . lett . *
90b * , 364 ( 1980 ) ; o. husser , in : proc .
7th int . conf . on polarization phenomena in nuclear physics ( paris , 1990 )
_ , nucl . phys . *
a460 * , 675 ( 1986 ) ; n. hintz _
_ , university of minnesota summary prog .
report ( 1984 - 1987 ) , p. 95 ; _ ibid .
_ prog . report ( 1989 ) , and l. ray , private communication ; g. w. hoffmann _
et al . _ ,
c * 21 * , 1488 ( 1980 ) , _ ibid .
_ c * 24 * , 541 ( 1981 ) ; g. d. alkhazov _ et al .
_ , akademiia nauk .
sssr , leningrad , institut iadernoi fiziki , no .
244 , 3 ( 1976 ) ; l. ray _
et al . _ , phys .
c * 23 * , 828 ( 1981 ) . | the nonrelativistic full - folding optical model approach for nucleon - nucleus scattering is extended into the relativistic regime . in doing so
, kinematical issues involving the off - shell lorentz boost of the colliding particles between the two nucleons and the projectile - nucleus center - of - mass reference frames have been taken into account .
the two - body effective interaction is obtained in the framework of the nuclear matter @xmath0-matrix using nucleon - nucleon optical model potentials that fully account for the inelasticities and isobar resonances in the continuum at nucleon energies up to 3 gev .
diverse nucleon - nucleon potential models were constructed by supplementing the basic paris , nijmegen , argonne or gelfand - levitan - marchenko inversion potentials with complex separable terms . in each case
the additional separable terms ensured that the combination led to nn scattering phase shifts in excellent agreement with experimental values . with each phase shift
fitting potential nuclear matter @xmath0-matrices have been formed and with each of those relativistic full - folding optical potentials for nucleon - nucleus elastic scattering determined .
application to such scattering for projectile energies to 1.5 gev have been made .
good and systematic agreement is obtained between the calculated and measured observables , both differential and integrated quantities , over the whole energy range of our study . a moderate sensitivity to off - shell effects in the differential scattering observables also is observed . |
for the successful implementation of new functionalities using the spin degree of freedom in technological platforms , experimental techniques are required for the generation of spin polarization , transport across relevant length scales and polarization detection . for device applications of non - equilibrium electron spins , triggered by either optical @xcite or electrical techniques @xcite , the generation and detection of spin currents
should be accomplished without the use of extremely strong magnetic fields and desirably using only electrical voltages . the electrical generation of spin polarization
have been widely studied in semiconductor structures of different dimensionality such as bulk samples @xcite and in n- and p - doped quantum wells ( qws ) @xcite .
the transport of spins polarized by an electrical current is of great interest not only due to the emergence of new phenomena but also due to possible applications in spin based electronics @xcite
. transverse transport of in - plane current - induced spin polarization ( cisp ) have been studied in l - shaped strained n - ingaas channels @xcite .
furthermore , it has been reported that the spins polarized in the out - of - plane direction by the spin hall effect ( she ) can be drive long distances in n - doped gaas epilayers @xcite . in this paper , we report on the electrical generation of spin polarization in two - dimensional electron gases under the action of spin - orbit fields and the spin transport over distances exceeding half millimetre away from current path in a transverse direction .
the structure discussed in this manuscript is gaas / algaas triple quantum well ( tqw ) , containg a dense two dimensional electron gas ( 2deg ) grown in the [ 001 ] direction .
the structure is symmetrically @xmath1-doped , with total density @xmath2 and low temperature mobility @xmath3 .
it contains a 26-nm - thick gaas central well with electron density of about @xmath4 and two 12-nm - thick lateral wells with approximately equal electron density of @xmath5 , each separated by 1.4-nm - thick al@xmath6ga@xmath7as barriers .
[ fig : lcisp ] ( a ) shows the calculated tqw band structure and subband charge density , where the black lines show the potential profile and the color lines show the first ( dark yellow ) , second ( orange ) and third ( violet ) occupied subbands . for the electrical generation of spin polarization
the sample was patterned into a six - contact hall bar geometry .
the main channel have length l = 500 @xmath8 between the side probes ( in the y - axis ) , width @xmath9 = 200 @xmath8 and four transverse channels of equal width 15 @xmath8 that extend out from the main channel ( in the x - axis ) as shown schematically in fig .
[ fig : lcisp ] ( b ) .
the current - induced spin polarization was measured by using the optical kerr rotation ( kr ) technique .
the spin polarization was generated by an electrical voltage modulated at @xmath10 = 1.1402 khz for lock - in detection .
a weak linearly polarized probe beam emitted from a pulsed ti : sapphire laser was scanned on the structure for the spatial detection of the induced spin polarization .
the polarization of the reflected beam was measured by a balanced photodetector and lock - in amplifier operating at @xmath11 .
the laser wavelength was tuned for maximum kr signal .
w.,scaledwidth=65.0% ] the sample was immersed in the variable temperature insert of a split coil superconductor magnet in the voigt geometry . in the longitudinal configuration ,
the kr was measured as a function of external magnetic fields while applying v@xmath12 at the central channel ( 1 - 2 ) [ see fig .
[ fig : lcisp ] ( b ) ] .
if the ensemble coherence time ( @xmath13 ) is longer than the pulse repetition ( with mhz frequency ) the signal does not fully decay from the previous pulse and overlaps with the signal generated by the following pulses . due to the constructive interference at certain resonant fields , the amplification of the precession ( with ghz frequency ) of those electrically polarized spin occurs and result in sharp resonance peaks .
these resonance peaks with spacing @xmath14 obey the periodic condition @xmath15 where @xmath16 is the laser repetition rate @xcite .
[ fig : lcisp ] ( c ) displays the rsa pattern as a function of the external magnetic field for several applied voltages .
the cisp amplitude increases with the increase of applied voltage and reaches its maximum value at 3 v. however , further increase of the applied voltage results in a strong reduction of the amplitude possibly due to heating effects @xcite .
[ fig : lcisp ] ( d ) displays the dependence of the cisp amplitude on the probe wavelength at a fixed voltage of 3 v. furthermore , the hole spin coherence time involved in the generation of the electron spin coherence , causes the reinforcement of the resonant peak amplitude at higher magnetic fields @xcite .
the experimental conditions for maximum signal amplitude , namely @xmath17 = 3 v and @xmath18 = 817 nm , were used in the following section .
now we switch from to the longitudinal transport to the transverse transport .
in contrast to the longitudinal transport , where the spin polarization was measured in a region where the local current flows , we measured the spin polarization in a region with minimal electric field .
for example , when @xmath17 is applied along the main channel ( y - direction ) the transverse transport will be into the side arms ( x - direction ) of the hall bar .
the reflectivity of the structure imaged as function of the position ( x , y ) is shown in fig .
[ fig : tcisp ] ( b ) .
the reflectivity map displays the central channel and the four transverse voltage probes .
the solid lines highlight the possible current paths and the red regions separate the conducting areas where the white lines are the edges of those regions . in fig .
[ fig : tcisp ] ( b ) , we imaged the kr signal as a function of b while applying @xmath19 in the x - axis using contacts 5 - 4 and scanning the probe spot at ( 0,y ) approximately 250 @xmath20 m away from the current path .
we observed the transverse drift of the cisp in a scan of about @xmath21y = 0.4 mm .
we also observed that the spacing @xmath22b between the rsa peaks varies with the position .
one can clearly see that the outer peaks are shifted towards higher fields for an increasing y - position indicating the variation of the @xmath0-factor as plotted in fig .
[ fig : tcisp ] ( b ) .
recently , the @xmath23-factor modification in a bulk ingaas epilayer by an in - plane electric field was studied @xcite .
they found a dependence of @xmath24 on the drift velocity : @xmath25 .
furthermore , the dependence of the electrical @xmath0 tensor variation on the magnetic field was also reported in quantum wells @xcite .
our data gives a @xmath0-factor variation in agreement with those reports . to extract the spin lifetime and polarization amplitude ,
the magnetic field scans were fitted using a lorentzian function @xmath26 $ ] with @xmath27 , where @xmath28 is the larmor frequency , @xmath29 is the reduced planck s constant , @xmath30 is the bohr magneton and b@xmath31 is the peak half - width .
the extracted parameters are plotted in fig .
[ fig : tcisp ] ( c ) and ( d ) .
surprisingly we found that in our structure the cisp transverse drift can drive a constant @xmath13 of about 6 ns by almost half millimeter .
the transport length of @xmath32= 0.625 @xmath33 0.058 mm was extracted from fitting the data in fig .
[ fig : tcisp ] ( d ) by @xmath34 .
those results of @xmath13 and @xmath35 are independent of the field chosen for analysis .
more detailed analysis , theoretical calculations and transport along x - axis are published in @xcite .
transverse configuration : ( a ) reflectivity map of the device .
( b ) kr as a function of b along ( 0,y ) mm axis with v@xmath12= 3 v. the symbol ( @xmath36 ) shows the spatial variation of electron @xmath23-factor . the extracted values of @xmath13 and amplitude are plotted in ( c ) and ( d ) respectively , the solid line is an exponential fit to the data .
t = 5 k.,scaledwidth=65.0% ]
we performed a detailed study on the cisp transport in a two - dimensional electron gas confined in gaas / algaas quantum wells .
we found long @xmath13 in the nanoseconds range and macroscopic drift distances in a direction transverse to the applied electric field .
additionally , a large spatial variation of the electron @xmath0-factor was observed .
f.g.g.h . acknowledges financial support from grant no .
2009/15007 - 5 , 2013/03450 - 7 , 2014/25981 - 7 and 2015/16191 - 5 of the so paulo research foundation ( fapesp ) .
s.u . acknowledges twas / cnpq for financial support .
9 ullah s , gusev g m , bakarov a k and hernandez f g g 2016 _ j. appl .
phys . _ * 119 * 215701 hernandez f g g , gusev g m and bakarov a k 2014 _ phys .
_ b * 90 * 041302(r ) kato y k , myers r c , gossard a c and awschalom d d 2004 _ phys .
lett . _ * 93 * 176601 ster n p , ghosh s , xiang g , zhu m , samarth n and awschalom d d 2006 _ phys .
lett . _ * 97 * 126603 sih v , myers r c , kato y k , lau w h , gossard a c and awschalom d d 2005 _ nat .
* 1 * 31 silov a , blajnov p , wolter j , hey r , ploog k and averkiev n 2005 _ semiconductors _ * 39 * 1323 aronov a g and lyanda - geller yu b 1989 _ jetp . lett . _ * 50 * 431 edelstein v m 1990 _ solid state commun . _
* 73 * 233 kato y k , myers r c , gossard a c and awschalom d d 2005 _ appl .
lett . _ * 87 * 022503 sih v , lau w h , myers r c , horowitz v r , gossard a c and awschalom d d 2006 _ phys .
* 97 * 096605 yugova i a , sokolova a a , yakovlev d r , greilich a , reuter d , wieck a d and bayer m 2009 _ phys .
lett . _ * 102 * 167402 luengo - kovac , macmahon m , huang s , goldman r s and sih v 2015 _ phys .
_ b * 91 * 201110(r ) chen y s , flt s , wegscheider and salis g 2014 _ phys .
_ b * 90 * 121304(r ) hernandez f g g , ullah s , ferreira g j , kawahala n m , gusev g m and bakarov a k 2016 _ phys .
_ b * 94 * 045305 | experimental studies of spin transport in a two - dimensional electron gas hosted by a triple gaas / algaas quantum well are reported . using time - resolved kerr rotation , we observed the precession of the spin polarization about a current - controlled spin - orbit magnetic field .
spatially - resolved imaging showed a large variation of the electron @xmath0-factor and the drift transport of coherent electron spins over distances exceeding half - millimetre in a direction transverse to the electric field . |
the financial support provided by the deutsche forschungsgemeinschaft is gratefully acknowledged . | we present an example revealing that the sign of the `` momentum '' @xmath0 of the wigner `` distribution '' function @xmath1 is not necessarily associated with the direction of motion in the real world .
this aspect , which is not related to the well known limitation of the wigner function that traces back to the heisenberg s uncertainty principle , is particularly relevant in transport studies , wherein it is helpful to distinguish between electrons flowing from electrodes into devices and vice versa .
recently , we critically analyzed @xcite the manner in which the wigner `` distribution '' function was used in studies on molecular transport relying upon finite isolated clusters @xcite .
these studies made us aware of a limitation of using the wigner function as a true momentum distribution function for transport , which we could not find in the literature and want to present here .
the wigner function @xmath1 is employed in many physical studies , including transport s , in spite of its physical limitations .
the limitation known from textbooks @xcite traces back to the heisenberg s uncertainty principle
. the wigner `` distribution '' function can be negative and should be not interpreted as a probability distribution , but rather as `` one step in the calculation never the last step , since '' is not measurable but `` is used to calculate other quantities that can be measured the particle density and current '' and ``
no problems are encountered as long as one avoids interpreting @xmath2 as a probability density '' ( quotations from ch .
3.7 , p. 203 of ref . ) . in transport , it is helpful to distinguish between incoming and outgoing electrons , i. e. , flowing from electrodes into devices and from devices into electrodes @xcite . to this aim
, it is necessary to use a physical property whose sign enables to indubitably assess that electrons are , say , left- or right - moving .
the averages of the particle ( probability ) current operator @xmath3 , \ ] ] or of the _ physical _ momentum @xmath4 do represent such properties .
above , @xmath5 and @xmath6 are annihilation and creation field operators for ( spinless ) electrons moving in one dimension , and @xmath7 stands for electron s mass . to see whether the momentum `` variable '' @xmath0 of the wigner function @xmath8 justifies to speak of left- or right - moving electrons depending on the sign of @xmath0 , let us consider the ground state @xmath9 of @xmath10 noninteracting electrons confined within a one - dimensional square well of width @xmath11 and infinite height .
electrons occupy energy levels @xmath12 , whose single - electron wave functions @xmath13 with @xmath14 , @xmath15 . in the ground state @xmath9 , the lowest @xmath10 levels are occupied up to the fermi `` momentum '' @xmath16 .
computing the wigner function of this system is straightforward @xmath17 see , e. g. , ref . ,
ch .
3.7 , pp .
202 - 203 .
( color online ) wigner function for two sizes @xmath11 computed at two points @xmath18 indicated in the legend .
gold s fermi wave vector @xmath19nm@xmath20 is used.,scaledwidth=45.0% ] one might think that one could use the wigner function as if it were a distribution function in cases where its shape resembles a fermi distribution . let us inspect the curves for @xmath1 computed as indicated above and presented in fig .
[ fig : wf ] .
in fact , at smaller sizes ( close to the linear size of the au@xmath21-clusters used in molecular wigner transport studies @xcite ) the wigner function does not bear much resemblance to a fermi function ( the lower curves of fig . [
fig : wf ] ) , at larger sizes ( much larger than those one could hope to tackle within ab initio calculations to correlated molecules , for which such a wigner - transport approach @xcite was conceived ) the curves ( the upper part of fig .
[ fig : wf ] ) become more similar to a step function , and one may think that this is encouraging . in reality , the contrary is true : as visible in fig .
[ fig : wf ] , _ mathematically _ one can calculate the wigner function for positive and negative `` momentum '' variables @xmath0 _
separately_. however , this mathematical separation does not reflect a physical reality : for any single - particle eigenstate @xmath22 the electron momentum vanishes @xmath23 left- and right - traveling waves are entangled with equal weight , and one can not speak of single - particle eigenstates representing left- or right - moving electrons only because wigner functions with positive or negative @xmath0-arguments can be computed .
this represents a further limitation of the usefulness of the wigner function , not related to the heisenberg s principle , which is particularly relevant for transport .
the current has a direction , and if one wants to unambiguously specify this direction , the wigner function @xmath1 is inappropriate ; a wigner function with negative ( positive ) `` momentum '' , @xmath24 ( @xmath25 ) , does not imply that left- and right - moving particles exist in the real physical world .
so , using @xmath26 and @xmath27 as if they were true momentum distributions of incoming electrons , as done in wigner approaches of molecular transport based on finite isolated clusters @xcite is not justified in quantum mechanics .
the above example demonstrates that , indeed , the textbook s warning mentioned in the beginning of this section is pertinent . |
starting with the seminal work of unruh @xcite , it has now been recognised for 30 years that a uniformly accelerated observer in minkowski space sees minkowski vacuum as a thermal state in temperature @xmath0 , where @xmath1 is the magnitude of the proper acceleration .
this result is of interest already in its own right within flat spacetime quantum field theory , and it has been confirmed by a number of methods @xcite . for relativists ,
the result is of particular interest because of its close mathematical similarity to the thermal properties of quantum fields in stationary black hole spacetimes @xcite .
a conceptually concrete way to address quantum effects in accelerated motion is to analyse a particle detector coupled to the quantum field . for the uniformly accelerated motion ,
a subtlety in such an analysis arises from the fact that the motion is _ stationary _ , that is , the orbit of a timelike killing vector . because of stationarity , the first - order perturbation theory transition probability over the whole trajectory is infinite , owing to the infinite total proper time .
however , this probability can be formally factorised into the product of the total proper time and a finite remainder , and the remainder can by stationarity be interpreted as the transition rate per unit proper time @xcite .
this regularisation prescription can be extended from stationary trajectories in minkowski space to curved spacetime @xcite , both for the unruh - dewitt monopole detector @xcite and a variety of its generalisations .
a recent review can be found in @xcite . for nonstationary motions the transition rate need
not be constant along the detector s trajectory , and a regularisation that relies on stationarity is no longer available .
the first message of this talk is : * for an unruh - dewitt monopole detector , the instantaneous transition rate is ill - defined without regularisation .
this observation appears to have been first made by schlicht @xcite , who showed that a conventional @xmath2 regularisation yields a lorentz - noninvariant transition rate for uniformly accelerated motion .
we discuss the mathematical reason for this phenomenon and show that the @xmath2 regularisation leads to a lorentz - noninvariant transition rate for _ every _ non - inertial trajectory .
the conventional @xmath2 prescription does therefore not provide a physically acceptable regularisation for the instantaneous transition rate .
schlicht proposed to regularise the transition rate in arbitrary motion by making the detector spatially extended in its instantaneous rest frame , with a spatial sensitivity profile that has a certain fixed shape but depends on a size parameter , and then letting the size parameter approach zero @xcite .
he showed that this regularisation yields the expected planckian spectrum for uniform acceleration , and he analysed a selection of nonstationary trajectories via mainly numerical methods .
we show that schlicht s regularisation yields a well - defined transition rate for every trajectory satisfying certain technical conditions , and we express the result as a manifestly finite integral formula that no longer involves regulators or limits . for the stationary trajectories the result agrees with that obtained in @xcite , and for nonstationary trajectories we extract asymptotic results that appear physically reasonable .
the second message of this talk thus is : * a spatial sensitivity profile is a viable regulator for the instantaneous transition rate .
the rest of the talk will put some flesh on these messages .
the main conclusions rely on a particular choice of the spatial profile function , but in section [ sec : shape ? ]
we present some evidence suggesting that that the zero size limit may be insensitive to the detailed form of the profile .
the talk is based on @xcite , where further detail can be found .
we work in four - dimensional minkowski spacetime with metric signature @xmath3 and in units in which @xmath4 .
boldface letters denote spatial three - vectors and sans - serif letters spacetime four - vectors , and a square of a spatial vector ( respectively spacetime vector ) is understood in the sense of the euclidean ( minkowskian ) scalar product .
consider a pointlike detector that moves in four - dimensional minkowski space on the world line @xmath5 , where @xmath6 is the proper time .
we take the detector to have two quantum states , denoted by @xmath7 and @xmath8 , which are eigenstates of the detector internal hamiltonian @xmath9 with the respective eigenvalues @xmath10 and @xmath11 , @xmath12 .
the detector is coupled to the real , massless scalar field @xmath13 with the interaction hamiltonian @xmath14 where @xmath15 is a coupling constant and @xmath16 is the detector s monopole moment operator , evolving in the heisenberg picture under @xmath9 .
@xmath17 is a switching function , which specifies how the interaction is switched on and off by an external agent .
suppose first that the switching function is smooth and has compact support , so that the initial and final states can be described in terms of the uncoupled system .
if before the interaction the field is in the minkowski vacuum @xmath18 and the detector in the state @xmath7 , the first - order perturbation theory probability of finding the detector in the state @xmath8 after the interaction is @xcite @xmath19 where the response function @xmath20 is given by @xmath21 and the correlation function @xmath22 is the pull - back of the wightman distribution , @xmath23 as @xmath22 is a well - defined distribution on @xmath24 @xcite , the transition probability given by ( [ eq : total - probability])([eq : w ] ) is well defined .
suppose then that no friendly neighbourhood external agent is available to switch the interaction off before we observe the detector .
we wish to ask : _ what is the probability of finding the detector in the state @xmath8 while the interaction is still switched on ? _ this is arguably the question encountered in a practical measurement where one looks at an ensemble of accelerated detectors ( say , atoms or ions ) at a given moment of time and counts what fraction of the detectors are in an excited state .
an attempt to answer this question within the detector model ( [ eq : hint ] ) would be to introduce in the switching function a sharp cutoff , @xmath25 , where @xmath6 is the proper time at which the detector is read and @xmath26 is the heaviside function . if we further push the switch - on to the asymptotic past , this would mean making in ( [ eq : total - probability])([eq : w ] ) the replacement @xmath27 .
formal manipulations then yield for the @xmath6-derivative of the response function the expression @xmath28 @xmath29 differs from the instantaneous transition rate only by a multiplicative constant that is independent of the trajectory , and we shall from now on suppress this constant .
the problem with these manipulations is that formula ( [ defexcitation - sharp - infty ] ) is ambiguous , since @xmath22 is a distribution with a singularity at the coincidence limit and the integration range has a sharp boundary at this singularity . to see that the problem is significant , suppose we go to a specific lorentz frame , @xmath30 ,
replace the wightman distribution by its conventional @xmath2 regularisation , @xmath31 and take the limit @xmath32 after performing the integral in ( [ defexcitation - sharp - infty ] ) .
assuming that the trajectory is sufficiently differentiable and has suitable falloff properties in the distant past , the result is @xcite @xmath33 \ , \label{resultado1-noninvariant}\end{aligned}\ ] ] where @xmath34 . the last term in ( [ resultado1-noninvariant ] )
vanishes for inertial trajectories but is lorentz - noninvariant wherever the proper acceleration is nonzero . in the usual distributional setting of integrating against smooth
test functions , the functions ( [ tradwightman - eps ] ) duly converge to the lorentz - invariant wightman distribution as @xmath35 @xcite , but the instantaneous transition rate ( [ defexcitation - sharp - infty ] ) falls outside this setting because of the sharp switch - off and retains a lorentz - noninvariant piece even in the limit @xmath36 . *
moral * : the instantaneous transition rate ( [ defexcitation - sharp - infty ] ) is ill - defined as it stands and needs to be regularised .
schlicht @xcite proposed to regularise the transition rate ( [ defexcitation - sharp - infty ] ) by giving the detector a spatial sensitivity profile that is rigid in the detector s instantaneous rest frame .
this idea can be motivated by the observation that real material systems ( say , atoms or ions ) are not pointlike .
technically , schlicht s proposal is to replace the field operator in the interaction hamiltonian ( [ eq : hint ] ) by a spatially smeared field operator , @xmath37 where @xmath38 are three unit vectors that together with the velocity @xmath39 form an orthonormal tetrad , fermi - walker transported along the trajectory .
the four quantities @xmath40 are thus fermi - walker coordinates in a neighbourhood of the trajectory @xcite .
the profile function @xmath41 is assumed to be non - negative and to integrate to unity , and @xmath42 is a positive parameter that determines the characteristic size of the smeared detector .
when @xmath43 is chosen to be the lorentzian function , @xmath44 schlicht showed that @xmath22 in ( [ defexcitation - sharp - infty ] ) gets replaced by @xmath45 ^ 2 } \ , \ ] ] where the unprimed and primed quantities are evaluated respectively at @xmath6 and @xmath46 .
note that @xmath47 ( [ corrschlicht ] ) is manifestly lorentz covariant .
schlicht further showed that the @xmath48 limit yields the planckian spectrum for the uniformly accelerated motion , thus agreeing with the regularisation that relies on stationarity @xcite .
he also examined the @xmath48 limit for a number of other trajectories , with physically reasonable results .
schlicht s results have been generalised by p. langlois @xcite to a variety of situations , including minkowski space in an arbitrary number of dimensions , quotients of minkowski space under discrete isometry groups , the massive scalar field , the massless dirac field and certain curved spacetimes .
langlois also observed that an alternative way to arrive at @xmath47 ( [ corrschlicht ] ) is to regularise the mode sum expression for the wightman function by an exponential frequency cut - off in the detector s instantaneous rest frame , rather than in a fixed lorentz frame .
when the regularised correlation function ( [ corrschlicht ] ) is substituted in ( [ defexcitation - sharp - infty ] ) , the existence of an @xmath32 limit is not obvious for an arbitrary trajectory since @xmath42 appears under the integral .
however , for trajectories that are sufficiently differentiable and have suitable falloff properties in the distant past , the limit exists and equals @xcite @xmath49 since the integrand in ( [ resultado1-infty ] ) remains finite at @xmath50 and since @xmath51 , formula ( [ resultado1-infty ] ) is manifestly well - defined .
formula ( [ resultado1-infty ] ) gives the transition rate as split into its odd and even parts in @xmath11 .
another useful split is into the inertial part and the noninertial correction , as introduced for stationary trajectories in @xcite .
this can be accomplished by a suitable addition and a subtraction in the integrand , with the result @xmath52 the first term in ( [ inermasacc ] ) is the transition rate of a detector in inertial motion , and the integral term is thus the correction due to acceleration . as the correction is even in @xmath11 , we see that the acceleration induces excitations and de - excitations with the same probability .
note that the correction term in ( [ inermasacc ] ) is nonvanishing for _ every _ noninertial trajectory .
note also that inversion of the cosine transform in ( [ inermasacc ] ) shows that @xmath53 fully determines @xmath54 as a function of @xmath55 and @xmath6 . from ( [ inermasacc ] )
it follows that @xmath56 has a large @xmath57 expansion that proceeds in inverse powers of @xmath58 , with coefficients given by @xmath6-derivatives of @xmath5 . in the leading order
we obtain @xmath59 which shows that for a generic trajectory the first correction to the inertial response is of order @xmath60 .
a case - by - case analysis of all stationary trajectories shows that the transition rate ( [ resultado1-infty ] ) for them agrees with that obtained with the regularisation that relies on stationarity @xcite .
in particular , in the special case of uniform acceleration of magnitude @xmath1 we have the planckian spectrum , @xmath61 as an example of nonstationary motion , consider a detector that moves in a timelike plane with the proper acceleration @xmath62 , where @xmath1 is a positive constant . in the distant past
the trajectory is asymptotically inertial , and we obtain the transition rate @xmath63 where the @xmath64-term holds uniformly in @xmath11 . in the distant future the trajectory has asymptotically uniform acceleration of magnitude @xmath1 , and we obtain the transition rate @xmath65 where @xmath66 stands for a term that goes to zero as @xmath67 . the first term in ( [ eq : as - unruh - future ] )
is the transition rate ( [ eq : planck ] ) in uniform acceleration .
the asymptotics thus agrees with what one would expect on physical grounds , both in the future and in the past .
the above results rely on the choice ( [ lorentzian ] ) for the profile function . while all sufficiently regular profile functions are known to yield the same @xmath48 limit for inertial motion @xcite , it is at present not known to what extent the @xmath48 limit might depend on the profile function for more general motions .
there is however a modified notion of spatial smearing in which we have been able to establish a result on profile - independence . for positive @xmath42 ,
the transition rate with this modified smearing reads @xmath68 where @xmath69 equations ( [ eq : smearedfdot - def ] ) and ( [ eq : g - def ] ) would follow from ( [ defexcitation - sharp - infty ] ) with the replacement ( [ eq : smeared - phi ] ) if it were known that the interchange of the @xmath70 and @xmath71 integrals is valid in a sense in which @xmath72 ( [ eq : g - def ] ) contains no distribution with support at @xmath73 . while we do not know whether the interchange can be justified in this sense
, we shall take equations ( [ eq : smearedfdot - def ] ) and ( [ eq : g - def ] ) as a _ definition _ of a detector model in their own right , arguing that this model captures at least some of the effects of the spatial smearing of section [ sec : profile ] .
now , if the trajectory is real analytic and satisfies suitable falloff conditions in the distant past , and if the profile function @xmath43 is smooth and has compact support , it can be shown @xcite that @xmath74 is well defined by ( [ eq : smearedfdot - def ] ) and ( [ eq : g - def ] ) for sufficiently small @xmath42 , and the @xmath48 limit exists and is given by ( [ resultado1-infty ] ) .
as this limit agrees with that obtained with the lorentzian profile function ( [ lorentzian ] ) ( which is not of compact support ) , we suspect that the equivalence of the two models of spatial smearing could be established for at least some classes of profile functions .
we have shown that regularising the transition rate of an accelerated unruh - dewitt detector on minkowski space by a spatial profile is a mathematically well - defined procedure and yields physically viable predictions in a number of situations .
for the lorentz - function spatial profile ( [ lorentzian ] ) the zero size limit could be computed explicitly , leading to the transition rate ( [ resultado1-infty ] ) . for other spatial profiles
the results remain to some extent inconclusive but they suggest that the zero - size limit may not be sensitive to the details of the profile .
we re - emphasise that the need for a spatial smearing arose because we chose to address the instantaneous transition rate _ while the interaction continues to be switched on _
, rather than the total excitation probability after the interaction has been smoothly switched on and off by an external agent .
it would be of interest to examine in comparison a pointlike detector whose smooth switching function is allowed to approach the step - function : might there exist limiting prescriptions that reproduce the effects of spatial smearing ? if the detector is turned on sharply at the finite proper time @xmath75 , the transition rate formula ( [ resultado1-infty ] ) is replaced by @xcite @xmath76 which is asymptotically proportional to @xmath77 as @xmath78 .
the total transition probability , obtained by integrating the transition rate ( [ resultado1-conc ] ) , is therefore infinite , owing to the violent switch - on event , regardless how small the coupling constant in the interaction hamiltonian is . for the stationary trajectories the transition rate ( [ resultado1-infty ] ) of a detector switched on in the asymptotic past
is constant in time , and the total transition probability is again infinite , now owing to the infinite amount of time elapsed in the past . in these situations
one may therefore have reason to view our results , all of which were obtained within first - order perturbation theory , as suspect .
however , in situations where the detector is switched on in the asymptotic past of infinite proper time and the total probability of excitation ( @xmath79 ) is finite , the first - order perturbation theory result should be reliable at least for the excitation rate , although the total probability of de - excitation ( @xmath80 ) then still diverges .
this situation occurs for the asymptotically inertial trajectory discussed in section [ sec : examples ] , and we expect it to occur whenever the proper acceleration vanishes sufficiently fast in the distant past
. it would be interesting to investigate to what extent our results can be generalised to the variety of situations to which schlicht s lorentzian profile detector was generalised in @xcite .
for example , do the formulas ( [ resultado1-infty ] ) and ( [ resultado1-conc ] ) generalise to spacetime dimensions other than four , and if yes , what is the form of the subtraction term ? does the clean separation of the spectrum into its even and odd parts continue ?
further , to what extent can the notion of spatial profile be employed to regularise the transition rate in a curved spacetime , presumably reproducing known results for stationary trajectories @xcite but also allowing nonstationary motion ?
in particular , might there be a connection with the regularisation prescriptions of the classical self - force problem @xcite ?
finally , would a nonperturbative treatment be feasible ?
jl thanks the organisers of the neb xii meeting for the invitation to present this work and their kind hospitality , pierre martinetti for discussions during the meeting and the british council for travel support .
jl acknowledges the hospitality and financial support of the isaac newton institute programme global problems in mathematical relativity " and the perimeter institute for theoretical physics and thanks howard j. magnuson for hospitality during the preparation of the manuscript . as was supported by an epsrc dorothy hodgkin research award to the university of nottingham .
99 poisson e 2005 in : _ proceedings of the 17th international conference on general relativity and gravitation _
( dublin , ireland , july 1823 , 2004 ) , edited by florides p , nolan b and ottewill a ( singapore : world scientific ) ( _ preprint _
gr - qc/0410127 ) | the instantaneous transition rate of an arbitrarily accelerated unruh - dewitt particle detector on four - dimensional minkowski space is ill defined without regularisation .
we show that schlicht s regularisation as the zero - size limit of a lorentz - function spatial profile yields a manifestly well - defined transition rate with physically reasonable asymptotic properties . in the special case of stationary trajectories , including uniform acceleration
, we recover the results that have been previously obtained by a regularisation that relies on the stationarity .
finally , we discuss evidence for the conjecture that the zero - size limit of the transition rate is independent of the detector profile . |
to model the dynamics of granular media a commonly used approach is the well - known discrete element method ( dem ) @xcite .
when applied to slow shearing @xcite the computation of frictional and therefore contact forces between particles may introduce numerical errors that can be larger than the precision of the integration scheme used in dem @xcite .
moreover , if irregular particles are considered typically polygons the definition of the contact plane between touching particles is not straightforward .
[ t ] particles ( green particles ) under shearing of top and bottom boundaries ( blue particles ) .
horizontally periodic boundary conditions are considered and a constant low shear rate is chosen . * ( b ) * illustration of two overlapping particles , where the overlap region @xmath0 between particles fully characterizes the contact force @xmath1.,width=529 ] in this paper we propose an approach to compute the tangential contact force with the same numerical precision as the normal force .
our improved procedure is specially suited for the case of slow shearing when large integration steps are needed for efficient computation , when studying e.g. the occurrence of avalanches @xcite and the emergence of ratcheting in cyclic loading @xcite .
in addition , we discuss the drawbacks of the common procedures for computing the contact plane between touching particles .
we will show that for polygonal ( anisotropic ) particles it varies discontinuously in time perturbing the proper convergence to the stationary state in stress controlled tests . to overcome this shortcoming we introduce a proper definition which does not depend on the shape of the touching particles and
is suited for both regular and irregular particles .
our model considers a two - dimensional system , as sketched in fig .
each particle has two linear and one rotational degree of freedom , being contained between two boundary plates shearing against each other @xcite .
the volumetric strain is suppressed , i.e. the position of the walls is fixed and there is no dilation .
periodic boundary conditions are imposed in the horizontal direction .
the evolution of the system is given by the integration of newton s equations of motion , where the resulting forces and momenta acting on each particle are given by the sum of all contact forces and torques applied on that particle , respectively .
the boundary particles move with a fixed shear rate .
the integration of newton s equation of motion is usually done with a predictor - corrector scheme @xcite , which consists of three main stages , namely prediction , evaluation and correction . in the prediction stage one extracts for each particle the predicted position and acceleration of the center of mass .
this is done by means of taylor expansions of the linear and angular positions , yielding the corresponding velocities and higher - order time derivatives , as functions of the current values @xcite . during the evaluation stage
, the predicted coordinate of each particle is used to determine the contact force @xmath2 at time @xmath3 .
since the method is not exact , there is a difference between the acceleration @xmath4 and the value obtained in the prediction stage , namely @xmath5 .
finally , the difference @xmath6 is used in the corrector step to correct the predicted position and its time derivatives , using proper weights for each time derivative @xcite .
the weights depend upon the order of the algorithm and the differential equation being solved .
these corrected values are used for the next integration step @xmath7 . in our simulations
we integrate equations of the form @xmath8 , using a fifth order predictor - corrector algorithm that has a numerical error proportional to @xmath9 for each integration step @xcite .
typically , the particles can neither break nor deform , i.e. fragmentation is neglected .
the deformation is usually modeled by letting particles overlap @xcite , as illustrated in fig .
the overlap between each pair of particles is considered to fully characterize the contact : the normal contact force is assumed to be proportional to the overlap area @xcite and its direction perpendicular to the contact plane , which is usually defined by the intersection points between the boundaries of the two particles @xcite .
the contact forces , @xmath1 , are decomposed into their elastic and viscous contributions , @xmath10 and @xmath11 respectively . the elastic part of the contact force is simply given by the sum of the normal and the tangential components , with respect to the contact plane , namely @xmath12 where the normal component reads @xmath13 with @xmath14 the normal stiffness , @xmath0 the overlap area and @xmath15 the characteristic length of the contact , which , for two particles @xmath16 and @xmath17 , is given by @xmath18 with @xmath19 and @xmath20 the area of the particle @xmath16 , and similarly for particle @xmath17 .
the tangential force is considered to be proportional to the elastic elongation @xmath21 of an imaginary spring also called cundall - stack spring @xcite at the contact , namely @xmath22 where @xmath23 is the tangential stiffness and @xmath21 is the elastic elongation updated as @xmath24 where @xmath25 is the time step of the dem simulation , and @xmath26 is the tangential component of the relative velocity @xmath27 at the contact point .
having both components of the elastic force , the tangential elastic elongation @xmath21 is updated according to the coulomb limit condition @xmath28 , with @xmath29 the inter - particle friction coefficient . if the coulomb condition is reached , particles are in the inelastic regime , and sliding is enforced by keeping constant the tangential force @xmath30 . here
the tangential elongation @xmath21 takes its maximum value @xmath31 .
otherwise , if the contact is in the elastic regime ( @xmath32 ) the elongation @xmath21 can increase in time , following eq .
( [ elasticelong ] ) .
the viscous force @xmath11 takes into account the dissipation at the contact which is important to maintain the numerical stability of the method .
this force is calculated as @xcite @xmath33 where @xmath34 is the reduced mass of the two touching particles and @xmath35 is the damping coefficient .
the suitable parameters for using this model are the interparticle friction @xmath29 , the normal stiffness @xmath14 , the ratio @xmath36 between tangential and normal stiffnesses and the ratio @xmath37 between tangential and normal restitution coefficients .
the restitution coefficients are defined from the contact stiffness and damping coefficient @xcite as @xcite @xmath38 for the normal component , where @xmath39 is the frequency of the damped oscillator , with @xmath40 the frequency of the elastic oscillator , @xmath34 the reduced mass , and @xmath41 is the effective viscosity .
we start in sec .
[ sec : sensitivity ] by describing the dependence of the above numerical procedure on the integration step , showing that the computation of the frictional forces yields a numerical error larger than the rest of the calculation . to overcome this shortcoming
we then describe in sec .
[ sec : geometrical ] a geometrical improvement whose associated error is , for any studied case , of the same order as the error of the predictor - corrector scheme . in sec . [
sec : contplane ] we address the particular case of irregular particles , discussing the most common ways of computing the contact plane between touching particles and proposing a more suitable definition . discussions and conclusions are given in sec .
[ sec : conclusions ] .
the entire algorithm above relies on a proper choice of the integration step @xmath25 , which should neither be too large to avoid divergence of the integration nor too small avoiding unreasonably long computational time . in this section
we explore the numerical error introduced by the frictional force in eq .
( [ elasticelong ] ) and determine the range of proper integration steps that guarantee convergence of the numerical scheme .
a typical criterion to chose the integration step @xmath25 is to take a value such that @xmath42 @xcite , where @xmath43 is the characteristic duration of a contact defined by @xcite , @xmath44 while in several cases , one uses an integration step much smaller than the thresholds above @xcite , for low shear rate , very small integration steps imply a high computational effort , and so @xmath25 should be chosen close to the threshold @xmath45 .
as shown below , the upper threshold ( fraction of @xmath43 ) below which the numerical scheme converges , strongly depends on ( i ) the accuracy of the approach used to calculate frictional forces between particles , ( ii ) on the corresponding duration of the contact and ( iii ) on the number of degrees of freedom . in a previous work @xcite
we have shown that considering values of @xmath25 close to the upper limit @xmath45 yields relaxation times depending on @xmath25 for the kinetic energy @xmath46 of each particle @xmath16 , when particles are subject to a non - zero friction .
in particular , in the absence of friction @xmath47 attains a stationary value , while when friction is present it changes monotonically .
remains fixed , while the particle at @xmath48 is initially touching particle @xmath16 .
the vector @xmath49 connecting the center of mass of particles @xmath16 and @xmath17 is initially oriented @xmath50 with respect to the @xmath51-axis . after applying the constant force @xmath52 to disc @xmath17 ,
the system relaxes till it reaches a new position ( dashed circumference ) . between its initial and final position particle
@xmath17 undergoes a displacement @xmath53 and a rotation @xmath54 ( see text).,width=321 ] to obtain the proper threshold as function of the parameters of our model , we consider the simple situation of two circular particles in contact , as sketched in fig .
[ fig2 ] , and study the kinetic energy of one of them under external forcing .
we start with two touching discs , say @xmath16 and @xmath17 , where @xmath16 remains fixed and @xmath17 is subject to a force @xmath52 perpendicular to its surface ( no external torque is induced ) along the @xmath51-axis . as a result of this external force ,
the disc @xmath17 undergoes translation and rotation .
the contact forces are obtained from the corresponding springs that are computed as described in sec .
[ sec : model ] and act against the external force .
this results in an oscillation of disc @xmath17 till relaxation ( dashed circle in fig .
[ fig2 ] ) with a final center of mass displacement of @xmath53 and a rotation around the center of mass of @xmath54 .
since @xmath52 is kept constant , the procedure is stress controlled .
plotting the kinetic energy as a function of time , yields an exponential decay as @xmath55 where @xmath56 is a relaxation time whose value clearly depends on the integration step @xmath25 .
typically @xcite the relaxation time @xmath56 depends on @xmath25 for integration steps larger than a given threshold @xmath57 .
we define this threshold as the proper upper limit for the suitable integration steps . in other words , we want to determine the upper bound @xmath58 where @xmath59 is a specific function that will be determined below .
( in units of @xmath43 ) as a function of * ( a ) * the integration step @xmath25 and * ( b ) * the normalized integration step @xmath60 , where the contact time @xmath43 is defined in eq .
( [ tc ] ) . here
the friction coefficient is kept fixed @xmath61 and different stiffnesses @xmath14 ( in units of n / m ) are considered .
the quotient @xmath60 collapses all the curves for different @xmath14 .
we find @xmath62 as illustrated in the inset ( see eq .
( [ tc ] ) ) . as
a final result one finds a constant @xmath63 . for other values of the friction coefficient one observes similar results . the relaxation time is also plotted as a function of the normalized integration step @xmath60 , when rotation is suppressed . *
( c ) * @xmath61 and different values of @xmath14 and for * ( d ) * @xmath64 and different values of @xmath29 .
the dashed horizontal line in ( b ) indicates the relaxation time of the kinetic energy in the absence of friction ( see text).,title="fig:",width=262 ] ( in units of @xmath43 ) as a function of * ( a ) * the integration step @xmath25 and * ( b ) * the normalized integration step @xmath60 , where the contact time @xmath43 is defined in eq .
( [ tc ] ) . here
the friction coefficient is kept fixed @xmath61 and different stiffnesses @xmath14 ( in units of n / m ) are considered .
the quotient @xmath60 collapses all the curves for different @xmath14 .
we find @xmath62 as illustrated in the inset ( see eq .
( [ tc ] ) ) . as
a final result one finds a constant @xmath63 . for other values of the friction coefficient one
observes similar results .
the relaxation time is also plotted as a function of the normalized integration step @xmath60 , when rotation is suppressed .
* ( c ) * @xmath61 and different values of @xmath14 and for * ( d ) * @xmath64 and different values of @xmath29 .
the dashed horizontal line in ( b ) indicates the relaxation time of the kinetic energy in the absence of friction ( see text).,title="fig:",width=266 ] we start by studying the influence of the stiffness , fixing @xmath61 and @xmath65 .
figure [ fig3]a shows the relaxation time @xmath56 of the kinetic energy of the two - particle system for @xmath66 and @xmath67 .
we see that decreasing @xmath25 , the relaxation time @xmath56 increases till it attains a maximum .
this is due to the computation of the tangential force using the cundall - spring scheme in eq .
( [ elasticelong ] ) that yields smaller values for smaller integration steps .
the stabilization of @xmath56 occurs when @xmath25 is small compared to the natural period @xmath68 of the system .
thus , @xmath69 is the largest value of @xmath25 for which we have this maximal relaxation time , in this case @xmath63 .
further , all the curves in [ fig3]a can be collapsed by using the normalized integration step @xmath60 , as shown in fig .
[ fig3]b . from the inset of fig .
[ fig3]b we also see that the contact time scales as @xmath62 , which comes from eq .
( [ tc ] ) .
( in units of @xmath43 ) as a function of the friction coefficient @xmath29 * ( a ) * when rotation is suppressed and * ( b ) * when rotation is considered . here ,
@xmath70 which corresponds to a contact time @xmath71 .
the normalized integration step @xmath72.,width=313 ] in the case where rotation is neglected , we obtain a constant @xmath73 as shown in fig . [ fig3]c .
this @xmath69 value is one order of magnitude smaller than the previous one in eq .
( [ aa ] ) and can be explained by considering an ` infinite ' friction coefficient in the rotational degree of freedom @xcite .
indeed , by comparing fig .
[ fig3]c with fig .
[ fig3]a , we see that the relaxation time @xmath56 is smaller when rotation is suppressed .
however , while fig .
[ fig3]c clearly shows that @xmath56 does not depend on the stiffness @xmath14 , the same is not true for the friction coefficient @xmath29 , as shown in fig .
[ fig3]d . in fig .
[ fig4]a we observe that , in the absence of rotation , there is a change of the relaxation time around @xmath74 , which is not observed when rotation is considered ( fig . [ fig4]b ) .
this transition occurs since for larger values of @xmath75 , one has @xmath76 , and therefore the frictional force increases and drives the system to relax faster .
further , with rotation one observes a larger @xmath56 because there is an additional degree of freedom , that also relaxes . of the tangential and normal stiffnesses .
* ( a ) * the convergence towards the relaxation time for decreasing integration steps using different stiffness ration . * ( b ) * the approximate logarithmic dependence of @xmath56 on @xmath36 ( see text ) .
here @xmath61 and @xmath65 .
no rotational constraint is here considered .
dependence of the relaxation time on the ration @xmath37 of the tangential and normal restitution coefficients .
* ( a ) * the convergence towards the relaxation time for decreasing integration steps using different restitution ratios .
* ( b ) * the relaxation time @xmath56 as a function of @xmath36 ( see text ) . here
@xmath61 and @xmath77 .
no rotational constraint is here considered.,width=377 ] we also check the dependence of the relaxation time on the two other parameters @xmath36 and @xmath37 .
figure [ fig5]a shows the relaxation time as a function of the normalized integration step @xmath60 , using different stiffness ratios and with fixed @xmath61 and @xmath37 .
we point out the following .
first the convergence to a stationary value of @xmath56 is faster for large stiffness ratio as @xmath78 , yielding larger values of @xmath69 .
second , the stationary value of @xmath56 decreases with the stiffness ratios .
this decrease is logarithmic , as shown in fig .
the dependence of @xmath56 on the stiffness ratio can be explained by taking into account that the larger the ratio @xmath36 the larger @xmath79 . since the tangential force @xmath79 controls the convergence of the numerical method , the larger @xmath36
the faster the dissipation and thus , the smaller the stationary value of @xmath56 .
the logarithmic dependence observed in fig .
[ fig5]b can be explained from direct inspection of eq .
( [ eq : res_coeff ] ) .
figure [ fig5]c , showing the relaxation time as a function of the ratio @xmath37 for fixed @xmath61 and @xmath77 can be explained similarly as in fig .
[ fig5]b . here
the dependence of the stationary value of @xmath56 on @xmath37 is approximately exponential . when the coulomb condition is fulfilled ( inelastic regime ) , the dependence on @xmath25 observed in all figures above , does not occur , since the strength of the tangential force is given by @xmath80 .
all the previous results are taken within the elastic regime .
this indicates that the improvements in the algorithm should be implemented when computing the elastic component of the tangential contact force , in eq .
( [ elasticelong ] ) , as explained in the next section .
using the cundall s spring @xcite , the relaxation time of the two particles only converges when @xmath25 is a small fraction @xmath69 of the contact time @xmath43 .
such dependence results from eq .
( [ elasticelong]),that includes @xmath25 in the computation of the tangential force in our predictor scheme which has an error of @xmath81 . to overcome this shortcoming , we propose a different expression to compute the elastic tangential elongation @xmath21 .
it is based on geometric relations and does not use @xmath25 .
our expression for @xmath21 contains only the quantities computed in the predictor step , guaranteeing a precision for @xmath21 of the same order as the one of the predictor - corrector scheme .
we illustrate our approach with the simple system of two discs , as sketched in fig .
[ fig2 ] , and discuss further the more realistic case of polygonal particles . in fig .
[ fig2 ] , the elastic elongation of the tangential spring results from the superposition of both translation and rotational degrees of freedom , @xmath82 . for the translational contribution
, the elastic elongation @xmath21 depends only on the relative position of the two particles . in this case
we substitute eq .
( [ elasticelong ] ) by the expression @xmath83 where @xmath84 and @xmath85 are the radii of the discs @xmath16 and @xmath17 respectively , @xmath49 is the vector joining both centers of mass and points in the direction @xmath86 ( see fig . [ fig2 ] ) .
index @xmath87 indicates quantities derived from the coordinates computed at the predictor step . considering only the rotational contribution ( @xmath49 constant )
, the elastic elongation @xmath21 depends only on rotation between times @xmath88 and @xmath7 : @xmath89 where @xmath90 is the angle of some reference point on particle @xmath17 at the predictor step of time @xmath88 .
( in units of @xmath43 ) when using the standard integration scheme ( squares ) and the proposed improved scheme ( circles ) . here ,
we perform the stress control test between two particles with irregular polygonal shape , as illustrated in fig .
[ fig1]b and rotation is neglected ( see text ) . *
( b ) * the relaxation time @xmath56 ( in units of @xmath43 ) using eqs .
( [ xi_trans ] ) and ( [ xi_rotation ] ) between two discs , as illustrated in fig .
[ fig2 ] ; for the three cases when considering only rotation , only translation or both , the relaxation time remains constant independent of the integration step ( see text).,width=313 ] using such expression , we have shown @xcite that the relaxation time @xmath56 is independent on the integration step .
this is due to the fact that all quantities in the expressions above have an error of the same order of the predictor - corrector scheme .
when considering polygonal particles , one must also take into account the shape of the particles . for polygons ,
the decomposition of the elastic elongation @xmath21 into translational and rotational contributions is not as trivial as for discs .
this stems from the fact that the contact point no longer lies on the vector connecting the centers of mass .
thus , one should recalculate each time the position of the center of mass ( only from translation ) and the relative position of the vertices ( only from rotation ) .
for that , instead of using eqs .
( [ xi_trans ] ) and ( [ xi_rotation ] ) , we compute the overlap areas between the two particles at times @xmath88 and @xmath7 and consider the dislocation of the corresponding geometrical centers .
this yields the translational contribution to the spring elongation .
the contribution from the particle rotation is computed by determining the angular change due to the rotation of the branch vector between times @xmath88 and @xmath7 .
figure [ fig6]a compares how the relaxation time varies with the normalized time step when the standard cundall approach is used ( squares ) and when our improved approach is introduced ( circles ) . clearly , the dependence on the integration step observed for the usual integration scheme disappears when our improved approach is introduced .
therefore , all the conclusions taken above for discs remain valid for polygons . in fig .
[ fig6]b we show the case of two discs for comparison .
is perpendicular to the plain containing the center of the discs , for polygons the same is not true , yielding the possibility of more than one suitable choice.,width=377 ]
an additional improvement concerning polygons is the proper choice of the contact plane .
usually shear system consider the simple situation of spherical particles . such particles yield a well defined contact plane , namely the vector perpendicular to the one connecting the centers of mass of the two touching particles .
figure [ fig7 ] illustrates this definition of contact plane for the case of a two - dimensional system . in the case of polygons
the branch vectors are not parallel and therefore no unique definition is possible .
usually @xcite one usually chooses the plane containing the intersection points and of the boundaries of the two touching particles .
when there are only two intersection points there is a unique possible contact plane .
however , due to the motion of the particles , the overlap between them may give rise to more than two intersecting points that yields different possible contact planes . in fig .
[ fig7 ] we illustrate a case where two different choices suit equally the condition above for a contact plane between two touching polygons . in this section
, we propose to define the contact plane as the line passing through the center of mass of the overlap area having an orientation defined by an angle @xmath91 ( with the @xmath51-axis ) given by the geometrical average of the orientation angles of each ( non - oriented ) edge of the overlap polygonal with respect to the @xmath51-axis .
as shown below such definition is unique and guarantees a continuous variation of the orientation of the contact plane while the polygons move , as it is the case in a model of irregular particles under shearing .
more precisely , for an overlap polygon delimited by @xmath92 edges , where each edge @xmath93 lies on a specific line @xmath94 , the contact plane in the two - dimensional case is defined from @xmath95 where @xmath96 with @xmath97 the coordinates of the center of mass of the overlap polygon , and @xmath98 with @xmath99 the length of segment @xmath100 . since the angles @xmath101 are angles between non - oriented lines , they can be acute or obtuse . to solve this indetermination we choose all the angles for the sum in eq .
( [ averlambda ] ) to be acute . then , after computing @xmath91 we choose either the line in eq .
( [ line ] ) or the one perpendicular to it , depending which one is more perpendicular to the line joining the center of mass of both particles . as a function of time when two polygons overlap each other .
two different definitions of contact plane are considered : the usual definition ( squares ) and the one proposed ( bullets ) , as given in eq .
( [ averlambda ] ) .
one sees that while the proposed definition varies continuously , the usual definition shows discontinuities that influence the contact between touching polygons ( see also figs .
[ fig9 ] and [ fig10]).,width=377 ] and * ( b ) * of @xmath91 for the usual definition of the contact plane .
the same evolution is plotted for * ( c)-(d ) * the new definition in eq .
( [ averlambda ] ) . in each case
two curves are plotted , corresponding to two different integration steps ( see text).,width=491 ] the main advantage of our definition in eq .
( [ averlambda ] ) is that while particles move the orientation of the contact plane varies continuously with their overlap ( polygonal ) area , as illustrated in fig .
[ fig8 ] with the curve marked by bullets . on the contrary , contact planes taken from two intersection points of the two polygons present frequent discontinuities , as can be seen from the examples with squares in fig .
such discontinuities occur since not only the size and orientation of the edges defining the overlap polygon vary in time , but also the _ number _ of edges .
there is also the advantage that such a definition enables the system to behave as physically expected .
for instance , during relaxation of a stress controlled test as illustrated in fig .
[ fig9 ] , the characteristic physical properties oscillate and converge to a stationary value . in fig .
[ fig9 ] we plot both @xmath102 and @xmath91 for both definitions .
we can see that the convergence of the new definition is similar to a damped oscillation , contrary to what is observed for the conventional definition .
such deviation from the damped oscillation behavior occurs due to the discontinuities observed in fig .
[ fig8 ] , which act as external excitations in the damping of the particles oscillation . as a consequence ,
the relaxation time associated with the discontinuous contact plane is smaller than when using our definition , as illustrated in fig .
[ fig10 ] .
we introduced a technique to improve the accuracy of the numerical scheme used to compute the evolution of particle systems , showing that the range of admissible integration steps has an upper limit significantly smaller than previously assumed .
the new approach for computing the frictional forces is an alternative to the cundall spring , given by eqs .
( [ xi_trans ] ) and ( [ xi_rotation ] ) and suits not only the simple situation of discs but also more realistic ones , where particles have polygonal shape .
further , our upper limit of the admissible range of integration steps was determined for a single contact using a stress controlled test .
its dependence on the stiffness , restitution coefficient and friction coefficient was carefully addressed .
when a larger system ( @xmath103 particles ) is studied , multiple contacts must be taken into account , which will yield an upper limit that we conjecture to be @xmath104 of the upper limit obtained above .
finally , for the case of polygonal particles , we also introduced a definition for the contact plane between two touching particles .
our definition is not only based on the intersections points of the particles in contact but also on the geometry of the overlap area .
thus , the contact plane varies continuously during the stress controlled tests and is unique , contrary to the usual definition .
the authors thank sean mcnamara and fernando alonso - marroqun for useful discussions .
we thank support by german - israeli foundation and by _ deutsche forschungsgemeinschaft _
, under the project he 2732781 .
pgl thanks support by _ deutsche forschungsgemeinschaft _
, under the project li 1599/1 - 1 .
hjh thanks the max planck prize .
s. luding , e. clment , a. blumen , j. rajchenbach , j. duran , phys .
e * 50 * , 4113 ( 1994 ) .
e. rougier , a. munjiza , n.w.m .
john , int .
j. numer .
eng . * 61 * , 856 ( 2004 ) .
cundall and o.d.l .
strack , gotechnique * 29 * , 47 - 65 ( 1979 ) . | we introduce two improvements in the numerical scheme to simulate collision and slow shearing of irregular particles .
first , we propose an alternative approach based on simple relations to compute the frictional contact forces .
the approach improves efficiency and accuracy of the discrete element method ( dem ) when modeling the dynamics of the granular packing .
we determine the proper upper limit for the integration step in the standard numerical scheme using a wide range of material parameters . to this end , we study the kinetic energy decay in a stress controlled test between two particles .
second , we show that the usual way of defining the contact plane between two polygonal particles is , in general , not unique which leads to discontinuities in the direction of the contact plane while particles move . to solve this drawback , we introduce an accurate definition for the contact plane based on the shape of the overlap area between touching particles , which evolves continuously in time . , , granular media , discrete element method , slow deformation , contact forces . 83.10.-y , 45.70.-n , 45.70.cc , 02.60.-x |
this work has been supported by rfbr grants 14 - 02 - 00914 , 14 - 02 - 31816 . | neutrino flavour oscillations in a nonuniformly moving matter are considered . the neutrino oscillation resonance condition in presence of matter , in the most general case when matter is moving with acceleration ,
is derived for the first time .
we predict that the effect of matter acceleration can have significant influence on neutrino oscillations pattern in different astrophysical environments .
in this short note we generalize our previous studies @xcite on neutrino flavour oscillations in an uniformly moving matter to the case when matter moves with acceleration .
consider two flavour ( electron and muon ) neutrino oscillations in nonuniformly moving matter .
the effective lagrangian of neutrino interactions with background matter can be expressed in the form @xmath0 where @xmath1 is the fermi constant , @xmath2 is the speed of matter , @xmath3 and @xmath4 and @xmath5 for the electron and muon neutrino correspondingly . here
@xmath6 is the electron number density in the reference frame for which the total speed of matter is zero .
note that we consider the matter composed of neutrons , protons and electrons .
obviously , neutrons and protons do not influence the oscillations . for the effective hamiltonian of neutrino flavour states evolution
we obtain @xmath7 where @xmath8 , @xmath9 is the vacuum mixing angle , @xmath10 is the neutrino energy and @xmath11 is the angle between the speed of matter and the direction of neutrino propagation ( it is supposed that the neutrino is propagating along @xmath12 direction ) . in the adiabatic approximation
the oscillation probability has the usual form @xmath13 , where the effective mixing angle @xmath14 and oscillation length @xmath15 are determined by the elements @xmath16 of the evolution hamiltonian ( [ h ] ) .
the straightforward calculations yields the neutrino flavour oscillations resonance condition @xcite @xmath17 in the considered case the neutrino oscillations probability gets its maximum value in a set of points @xmath18 that are the solutions of eq .
( [ res_cond ] ) . note that in general case eq .
( [ res_cond ] ) is not linear in respect to @xmath12 . obviously , that in case of monotonic @xmath12 dependence of the density @xmath19 there can be only one resonance point
.
consider matter motion with a constant acceleration @xmath20 .
then the matter velocity is given by @xmath21 where @xmath22 is the initial matter speed .
in this case one can obtain the effective electron matter density in the form @xmath23 in fig .
[ fig01 ] the function @xmath24 is plotted .
it follows that @xmath19 significantly increases in case of matter motion against neutrino propagation ( @xmath25 ) and almost vanishes in the opposite case ( @xmath26 ) . in the nonrelativistic and ultrarelativistic limits of matter motion eq .
( [ ne_v ] ) can be simplified as follows @xmath27 in the corresponding evaluation it is supposed that @xmath28 . when nonrelativistic matter is moving with a constant acceleration the corresponding shift of the electron number density
is given by @xmath29 , where @xmath30 and @xmath31 .
for instance , during a supernova core - collapse @xmath32 is up to 0,2 that yields an order of @xmath33 shift to the effective resonance number density @xmath19
. it would be interesting to consider possibilities of observing this effect . and the angle of propagation @xmath11 .
the red arrows defines the direction of a neutrino flux .
the left and right plots corresponds to @xmath28 and @xmath34 accordingly.,title="fig : " ] and the angle of propagation @xmath11 .
the red arrows defines the direction of a neutrino flux .
the left and right plots corresponds to @xmath28 and @xmath34 accordingly.,title="fig : " ] |
numerous observations on neutrinos from the sun@xcite , the atmosphere@xcite , the reactor@xcite , and the accelerator@xcite suggest that neutrinos are massive and hence there are mixings in the lepton sector .
this fact means that the standard model ( sm ) has to be extended so that the neutrino masses and the lepton mixings are introduced into the model .
lots of models to explain those experimental results have been proposed . among them
, a model with the seesaw mechanism@xcite has a promising attribute , in which tiny neutrino masses are naturally induced .
neutrino experiments also have revealed that the mixings in the lepton sector are much larger than those in the quark sector .
this fact may imply that the lepton flavor number is strongly violated in the physics beyond the sm .
therefore , we can expect that the nature might exhibit sizable lepton flavor violation ( lfv ) and hence we could observe the remains of physics at the high energy scale . in the minimal supersymmetric standard model with heavy right - handed neutrinos ( mssmrn ) , in which the seesaw mechanism is realized , the lfv with charged leptons ( clfv ) is expected to become large@xcite . in this class of models ,
the renormalization effect due to the neutrino yukawa couplings induces a significant size of off - diagonal elements of the slepton mass matrix , @xmath1 ( @xmath2 , @xmath3 ) , which are the seeds of the clfv . here ,
the flavor indices @xmath4 and @xmath5 should be understood to also indicate the mass eigenstates of the charged lepton fields .
concretely , the superpotential with the neutrino yukawa couplings @xmath6 and the majorana masses for the right - handed neutrinos @xmath7 includes the following terms , @xmath8 where @xmath9 , @xmath10 , and @xmath11 denote the chiral supermultiplet for the right - handed neutrinos , that for the lepton doublets , and that for the up - type higgs field , and @xmath12 is the anti - symmetric tensor for @xmath13 fundamental representation . the indices @xmath14 and @xmath15 are for the generation of right - handed neutrinos , which do not necessarily indicate the mass eigenstates . the renormalization group equation for the off - diagonal elements of the slepton mass matrix is given as@xcite , @xmath16_\alpha^{\ \beta } , \quad ( \alpha \neq \beta ) , \label{eq : rgeforl}\end{aligned}\ ] ] where @xmath17 is the soft supersymmetry ( susy ) breaking mass matrix for the right - handed sneutrino , and @xmath18 is that for the up - type higgs doublet .
the matrix @xmath19 denotes the tri - linear scalar couplings corresponding to the first term in eq . .
note that if the neutrino yukawa couplings do not exist , then there is no lfv effect .
we can approximately solve eq .
as @xmath20 with a cutoff scale @xmath21 and a typical mass scale for right - handed neutrinos @xmath22 . here , the universal soft susy breaking at @xmath23 is assumed , and @xmath24 is the parameter for sfermion masses and @xmath25 is that for scalar tri - linear couplings .
= 0.9 cm ( 10,7 ) ( 0.5,4)(1,0)2 ( 1.5,4)(1,0)0 ( 0.6,4.2)@xmath26 ( 2.5,4 ) ( 2.5,4)(0.5,0)8(1,0)0.3 ( 3.5,4)(1,0)0 ( 3.2,4.2)@xmath27 ( 4,3.2)@xmath28 ( 4.5,4 ) ( 5.5,4)(1,0)0 ( 5.2,4.2)@xmath29 ( 6.5,4 ) ( 6.5,4)(1,0)2 ( 8,4)(1,0)0 ( 8,4.2)@xmath30 ( 4.5,4 ) ( -2,0)(-2.01,0.35)(-1.88,0.68 ) ( -1.88,0.68)(-1.78,1.03)(-1.53,1.29 ) ( -1.53,1.29)(-1.3,1.55)(-1,1.73 ) ( -1,1.73)(-0.69,1.9)(-0.35,1.97 ) ( -0.35,1.97)(0,2.03)(0.35,1.97 ) ( 2,0)(2.01,0.35)(1.88,0.68 ) ( 1.88,0.68)(1.78,1.03)(1.53,1.29 ) ( 1.53,1.29)(1.3,1.55)(1,1.73 ) ( 1,1.73)(0.69,1.9)(0.35,1.97 ) ( 4.75,6)(1,0)0 ( 4.5,4 ) ( -2,0)(-2.36,0.417)(-1.88,0.68 ) ( -1.88,0.68)(-1.47,0.85)(-1.53,1.29 ) ( -1.53,1.29)(-1.54,1.84)(-1,1.73 ) ( -1,1.73)(-0.58,1.6)(-0.35,1.97 ) ( -0.35,1.97)(0,2.35)(0.35,1.97 ) ( 2,0)(2.36,0.417)(1.88,0.68 ) ( 1.88,0.68)(1.47,0.85)(1.53,1.29 ) ( 1.53,1.29)(1.54,1.84)(1,1.73 ) ( 1,1.73)(0.58,1.6)(0.35,1.97 ) ( 4.4,6.5)@xmath31 ( 2.5,-1.2 ) ( 6.2,1.8)@xmath32 in terms of the mass insertion method , we can see that the off - diagonal elements of the slepton mass matrix are the origins of clfv .
this is diagrammatically shown in fig .
[ fig : chargedlfv ] . from this diagram , it is obvious that the element @xmath33 is relevant to the process @xmath34 . in this class of models , the off - diagonal elements can become large so that the typical values of predicted branching ratios are within a sensitivity reach of near future experiments@xcite .
therefore , the search for the clfv process is one of the promising ways to inspect the new physics effect beyond the sm . in this article , however , we investigate an alternative approach to explore the lfv , the search for the processes of the lfv with neutrinos ( nlfv ) at a long baseline ( lbl ) neutrino oscillation experiment . in the forthcoming experiments ,
the oscillation parameters such as the mixing angles and the squared mass differences are expected to be determined with high precision@xcite .
therefore , the measurement of nlfv effects might be possible .
the feasibility studies on the nlfv interaction search at future lbl experiments without assuming a specific model are made by refs .
the current experimental bounds on nlfv interactions are given in ref .
the sensitivity for nlfv effects to solar neutrinos@xcite , atmospheric neutrinos@xcite , the lsnd results@xcite and supernova neutrinos@xcite has also been considered .
it is pointed out that the nlfv signal is enhanced by the interference effect between the amplitude including the nlfv interactions and that of the standard oscillation ( so ) .
we here investigate the nlfv interactions in the mssmrn .
the origins of nlfv processes are the same as those of clfv processes , which are the off - diagonal elements of the slepton mass matrix .
they can become significant in this framework .
in addition , there is an enhancement mechanism due to the interference effect .
it can be expected that the detectable magnitude of nlfv effects is induced .
we make the numerical calculations of the size of the nlfv couplings and show the correlation between the nlfv and the clfv . in sec .
[ sec : paramtrization ] , we recapitulate the model independent approach in the detection of nlfv effects at lbl neutrino oscillation experiments
. we also show the way to parameterize the nlfv interactions . in sec .
[ sec : mssm ] , we calculate these nlfv couplings in the mssmrn and numerically evaluate the size of them under the universal soft susy breaking scenario , so - called the constrained mssmrn . here , we concentrate our attention on the nlfv interactions which are relevant in the oscillation channel @xmath35 . finally , in sec . [
sec : summary ] we will give a summary .
in appendix [ app : model ] , we describe the model in order to make our notation clear , and in appendix [ app : calculation ] we give formulae of the nlfv interactions .
in this section , we explain how to parameterize the new physics effect with the model independent way@xcite .
first , we note that in neutrino oscillation experiments we do not observe neutrinos themselves but do their products , corresponding charged leptons . propagating neutrinos
appear only in intermediate states .
therefore , the existence of the nlfv interactions suggests that there are some amplitudes whose initial and final states are the same as those of the so which means the neutrino oscillation with sm interactions . to make the argument clear , we show an example . when we assume @xmath35 oscillation measurement at a neutrino factory experiment , all we can know are the facts , the decay of muons at a muon storage ring and the appearance of tau leptons in a detector which is located at a length @xmath36 away from the source of the neutrino beam just after the time @xmath37 , where @xmath38 is speed of light .
we interpret these events as the evidence of the so , @xmath39 .
the amplitude for this process @xmath40 can be expressed by the product of the amplitudes for the sub - processes ; @xmath41{\text{osc . } } \nu_{\tau } ) a_d(\nu_{\tau } d \rightarrow \tau^{- } u ) , \label{eq : smamplitude}\end{aligned}\ ] ] where @xmath42 ( @xmath43 ) denotes all the other particles than a muon ( tau lepton ) in the initial ( final ) state , which can be measured in principle . in this example , @xmath42 is @xmath44 ( a down - type quark ) in @xmath45 , and @xmath43 means @xmath46 and @xmath47 in @xmath48 and @xmath49 ( a up - type quark ) in @xmath45 .
the subscripts , @xmath50 , and @xmath44 , indicate `` at the _ source _ of the neutrino beam '' , `` at the _ propagation _ process '' , and `` at the _ detection _ process '' , respectively .
suppose that there is an effective four - fermi nlfv interaction , @xmath51 we have the same signal through the other process than eq . ;
@xmath52{\text{no osc . } } \nu_{\tau } ) a_{d}(\nu_{\tau } d \rightarrow \tau^{- } u ) .
\label{eq : nlfvamplitude}\end{aligned}\ ] ] the external particles in eq .
are completely the same as those in eq .
( [ eq : smamplitude ] ) .
therefore , we can not distinguish the contributions from these two amplitudes . in quantum mechanics ,
we first sum up these amplitudes and next square the summation in order to obtain the transition rate .
therefore , an interference term arises between these two amplitudes for this process : @xmath53 + |a_{\rm nlfv}|^{2 } .
\label{eq : interference}\end{aligned}\ ] ] the term of the so , the first term of the right - hand side , gives the leading contribution , and it is described by using the muon decay width @xmath54 @xmath55 and the cross - section for the charged current interaction @xmath56 @xmath57 as @xmath58 where @xmath59 is the oscillation probability for @xmath35 in the so , which is defined as @xmath60{\text{osc . } } \nu_{\tau})|^{2}$ ] .
the second term which represents the interference between the amplitude of the so eq . ( [ eq : smamplitude ] ) and that including the nlfv interaction eq .
( [ eq : nlfvamplitude ] ) : @xmath61 = \gamma \times 2 \text{re}\left [ \frac{\lambda}{2 \sqrt{2 } g_{f } } a_{p}^{*}(\nu_{\mu}\xrightarrow[]{\text{osc . } } \nu_{\tau } ) a_{p}(\nu_{\tau } \xrightarrow[]{\text{no osc . } } \nu_{\tau } ) \right ] \times \sigma,\end{aligned}\ ] ] where @xmath62 is the fermi constant .
note that the nlfv effect contributes to the oscillation probability not quadratically but linearly .
thus the effect can be _ enhanced _ and hence even if the nlfv coupling is small , it can contribute the oscillation probability significantly@xcite .
we now turn to the parametrization of effective couplings for nlfv interactions .
as we have already shown , the amplitude for _ the neutrino oscillation process _ can be divided into three pieces , @xmath48 , @xmath63 , and @xmath64 .
first , we consider the decay process of parent particles , @xmath48 .
since all final states must be the same , nlfv interactions with @xmath65 type are important for the neutrino factory experiment@xcite .
we can introduce the interference effect by treating the initial state of a propagating neutrino as the superposition of all flavor eigenstates .
for the case of eq . , we can take the initial neutrino state @xmath66 as @xmath67 where @xmath68 .
it can be generalized to the case of an initial neutrino with any flavor by using the source state notation which is introduced in ref .
@xcite as is not necessarily unitary .
] @xmath69 we can include the total nlfv effect into the oscillation probability as @xmath70 with the propagation hamiltonian for the so @xmath71 ; @xmath72 where @xmath73 is the neutrino energy , @xmath74 is the usual matter effect which is given as @xmath75 by using the electron number density @xmath76 , @xmath77 is the mixing matrix for the lepton sector , and @xmath78 ( @xmath79 ) is the mass squared difference for the solar ( atmospheric ) neutrino oscillation .
next , we consider the propagation process , @xmath63 .
the nlfv interactions modify the hamiltonian for neutrino propagation as@xcite , @xmath80 where @xmath81 is the extra matter effect due to nlfv interactions , which is defined as @xmath82 where @xmath83 is the number density for the particle @xmath84 . assuming the matter which consists of the same number of electron , neutron and proton
, we can reduce it to @xmath85 with @xmath86 note that to consider the magnitude of the matter effect , the type of the interaction is irrelevant since matter particles are at rest and hence the dependence on their chirality is averaged out@xcite .
then , we make a comment on nlfv interactions which affect a detection process , @xmath45 .
we can adopt a quite similar treatment to that at the source of the neutrino beam . in this article
, we consider the case in which the nlfv interactions do not depend on the flavor of target - quark , which is almost the case for the so - called constrained mssmrn .
therefore , we have the neutrino state for the detection process in the following form , @xmath87 finally , the transition probability including the whole nlfv effects is given as @xmath88 where @xmath89
in this section , we evaluate the effective couplings for nlfv interactions in the mssmrn .
the nlfv interactions are induced by the off - diagonal elements of the slepton mass matrix , which are similar to the clfv interactions .
we can naively estimate the size of the nlfv parameter @xmath90 from the diagram which is shown in fig . [ fig : diagram - for - epsilon ] as@xcite @xmath91 here @xmath92 is the gauge coupling for @xmath13 and @xmath93 is the typical susy breaking scale .
= 0.9 cm ( 10,7 ) ( 0.5,4)(1,0)2 ( 1.5,4)(1,0)0 ( 0.5,4.3)@xmath26 ( 2.5,4 ) ( 2.5,4)(0.5,0)8(1,0)0.3 ( 3.5,4)(1,0)0 ( 3.1,4.3)@xmath27 ( 4.5,4.3)@xmath29 ( 5.4,4 ) ( 3.8,4 ) ( 6,4)(1,0)0 ( 4.7,4)(1,0)0 ( 5.7,4.3)@xmath94 ( 6.5,4 ) ( 6.5,4)(1,0)2 ( 8,4)(1,0)0 ( 8,4.3)@xmath95 ( 4.5,4 ) ( -2,0)(-2.01,0.35)(-1.88,0.68 ) ( -1.88,0.68)(-1.78,1.03)(-1.53,1.29 ) ( -1.53,1.29)(-1.3,1.55)(-1,1.73 ) ( -1,1.73)(-0.69,1.9)(-0.35,1.97 ) ( -0.35,1.97)(0,2.03)(0.35,1.97 ) ( 2,0)(2.01,0.35)(1.88,0.68 ) ( 1.88,0.68)(1.78,1.03)(1.53,1.29 ) ( 1.53,1.29)(1.3,1.55)(1,1.73 ) ( 1,1.73)(0.69,1.9)(0.35,1.97 ) ( 4.8,6)(1,0)0 ( 4.5,4 ) ( -2,0)(-2.36,0.417)(-1.88,0.68 ) ( -1.88,0.68)(-1.47,0.85)(-1.53,1.29 ) ( -1.53,1.29)(-1.54,1.84)(-1,1.73 ) ( -1,1.73)(-0.58,1.6)(-0.35,1.97 ) ( -0.35,1.97)(0,2.35)(0.35,1.97 ) ( 2,0)(2.36,0.417)(1.88,0.68 ) ( 1.88,0.68)(1.47,0.85)(1.53,1.29 ) ( 1.53,1.29)(1.54,1.84)(1,1.73 ) ( 1,1.73)(0.58,1.6)(0.35,1.97 ) ( 4.3,6.5)@xmath31 ( 4.3,2)@xmath96 ( 5.4,1.4)(1,0)2.5 ( 5.4,1.4)(4,-1)2.5 ( 5.4,1.4 ) ( 6.9,1.4)(-1,0)0 ( 6.92,1.02)(4,-1)0 ( 7.6,1.6)@xmath46 ( 6.9,0.6)@xmath97 ( 1.5,3.2)@xmath28 this relation means that there is a correlation between two processes , @xmath98 and @xmath99 .
we here concentrate our attention on the nlfv associated with the tau lepton because that in @xmath100-@xmath97 sector is strongly constrained by corresponding clfv processes .
the current experimental bound on the branching ratio of @xmath101 is @xmath102 at 90% confidence level@xcite , and then this experimental bound constrains the nlfv coupling parameter @xmath103 . according to this naive estimation , the value of @xmath103 may become as large as @xmath104 . such size of nlfv interactions would be detected at a future lbl experiment such as the neutrino factory@xcite . in this subsection , we explain the calculation to obtain @xmath106 in detail and compare them with that of clfv processes .
the thorough results are given in appendix [ app : calculation ] .
for example , the effective lagrangian relevant with the nlfv interactions which give potentially significant contribution to the oscillation @xmath107 in a neutrino factory is given as@xcite @xmath108 these effective couplings arise from penguin - type diagrams and from box - type diagrams as shown in appendix [ app : calculation ] .
the calculation of box diagrams is straightforward .
it is almost the same as that of clfv processes , e.g. , @xmath100-@xmath97 conversion@xcite except for the fact that only the @xmath65 type interactions are taken into account in the calculation of @xmath109 and @xmath110 .
however , it is necessary to make an attentive calculation for penguin - type diagrams .
in general , while the neutral and electromagnetic currents corresponding to fig . [
fig : chargedlfv ] take the following form , @xmath111 f_\beta ( p ) \nonumber \\ & + { \rm h.c .
} , \end{aligned}\ ] ] the charged current corresponding to fig .
[ fig : diagram - for - epsilon ] is decomposed to @xmath112 here , @xmath113 denotes the charged lepton field @xmath114 or the neutrino field @xmath115 , @xmath84 is the momentum for the incoming particle , and @xmath116 is that for the gauge boson .
all the coefficients , @xmath117 ( @xmath118 ) , @xmath119 , @xmath120 ( @xmath121 ) , @xmath122 , @xmath123 ( @xmath124 ) , and @xmath125 , are the functions of @xmath126 . in the limit @xmath127 , @xmath120 and @xmath122 for the electromagnetic current must vanish due to the gauge symmetry @xmath128 . on the other hand , those for the neutral and the charged current do not vanish because the corresponding gauge symmetry @xmath13 is broken .
in other words , the lorentz structure of the nlfv interaction which is shown in fig .
[ fig : diagram - for - epsilon ] is different from that of the clfv process @xmath34 with a real photon emission . the former one is dominated by the vector exchange interaction ( @xmath129 ) , and @xmath96 and @xmath126 is of @xmath130 , essentially we can put @xmath131 . ] while the latter is done by a di - pole type interaction ( @xmath132 ) .
therefore , they are definitely correlated each other but are not the same functions .
= 0.8 cm ( 10,7 ) ( 1.8,4)(2,0.5 ) ( -0.5,4)(1,0)3 ( 0.5,4)(1,0)0 ( -0.7,4.3)@xmath26 ( -0.7,3.5)@xmath133 ( -0.3,3.6)(1,0)0.8 ( 1.3,4.5 ) @xmath134 ( 2,4)(1,0)0 ( 1.8,3.6)(1,0)0.8 ( 2,3.1)@xmath135 ( 2.5,4 ) ( 2.5,4)(0.5,0)8(1,0)0.3 ( 4.65,4)(1,0)0 ( 4.1,4.3 ) @xmath136 ( 4.1,3.6)(1,0)0.8 ( 3.6,3.1)@xmath137 ( 1.15,4 ) ( 6.5,4 ) ( 6.5,4)(1,0)2 ( 8,4.3)@xmath138 ( 6.8,3.6)(1,0)0.8 ( 7.8,3.5)@xmath135 ( 8,4)(1,0)0 ( 4.5,4 ) ( -2,0)(-2.01,0.35)(-1.88,0.68 ) ( -1.88,0.68)(-1.78,1.03)(-1.53,1.29 ) ( -1.53,1.29)(-1.3,1.55)(-1,1.73 ) ( -1,1.73)(-0.69,1.9)(-0.35,1.97 ) ( -0.35,1.97)(0,2.03)(0.35,1.97 ) ( 2,0)(2.01,0.35)(1.88,0.68 ) ( 1.88,0.68)(1.78,1.03)(1.53,1.29 ) ( 1.53,1.29)(1.3,1.55)(1,1.73 ) ( 1,1.73)(0.69,1.9)(0.35,1.97 ) ( 4.8,6)(1,0)0 ( 4.5,4 ) ( -2,0)(-2.36,0.417)(-1.88,0.68 ) ( -1.88,0.68)(-1.47,0.85)(-1.53,1.29 ) ( -1.53,1.29)(-1.54,1.84)(-1,1.73 ) ( -1,1.73)(-0.58,1.6)(-0.35,1.97 ) ( -0.35,1.97)(0,2.35)(0.35,1.97 ) ( 2,0)(2.36,0.417)(1.88,0.68 ) ( 1.88,0.68)(1.47,0.85)(1.53,1.29 ) ( 1.53,1.29)(1.54,1.84)(1,1.73 ) ( 1,1.73)(0.58,1.6)(0.35,1.97 ) ( 4.4,6.5)@xmath139 ( 4.5,3.5)(-0.35,1.97)(0,2.03)(0.35,1.97 ) ( 3.95,5.4)(-4,-1)0 ( 4.4,5)@xmath140 ( 0.8,2)@xmath96 ( -0.5,2.5)(3,-4)0.6 ( -0.9,2)@xmath116 in the calculation of nlfv diagrams , we have to pay attention to the following two facts : ( i ) we regard neutrinos as highly off - shell particle , and ( ii ) we must avoid counting one contribution twice .
first , we explain the reason why neutrinos behave as highly off - shell particles , and hence it is necessary to treat that neutrinos are massless in diagrams for effective nlfv interactions . in fig .
[ fig : offshell ] , from the viewpoint of the uncertainty principle , the condition for neutrino oscillation to occur is described as@xcite @xmath141 where @xmath142 is the uncertainty of the position and the time in which neutrinos are produced , @xmath143 is the uncertainty of the energy - momentum of the outgoing neutrino , and @xmath144 is the neutrino mass squared difference corresponding to the lbl oscillation experiment .
the inequality results from the fact that the neutrino production position must be determined much more accurately than the baseline length @xmath36 .
the fact that the neutrino oscillation phenomena are observed in the lbl experiment requires that the equality at the most right - hand side should be satisfied .
thus , the uncertainty of the squared momentum must conform to the following relation @xmath145 where @xmath146 denotes the average of @xmath147 .
this inequation shows that the average of the neutrino momentum is much larger than its mass , and it follows from this that neutrinos are generally highly off - shell fields in oscillation experiments .
this also means that all the diagrams for the nlfv interactions include the off - shell neutrino as external lines . in the exact meaning of the field theory
, we do not have the method to calculate diagrams with off - shell external legs .
however , we can evaluate such diagrams by treating as if the neutrinos were massless .
for the practical purpose , we make the following replacement which we refer as the off - shell prescription in calculations of the type of diagram shown in fig . [
fig : offshell ] ; for the flavor eigenstate@xcite appears in eq . .
however , we finally neglect it in our off - shell prescription . ]
@xmath148 next , we turn to the problem of the double counting and show the way to solve it .
first , we note that we must calculate the process @xmath149 with the method of the field theory , so that we must not calculate the nlfv effect for each stage because we can not observe the each stage separately .
= 0.6 cm ( 10,7 ) ( -0.5,4)(1,0)3 ( 0.5,4)(1,0)0 ( -0.7,4.3)@xmath150 ( 1.1,4.3)@xmath151 ( 2,4)(1,0)0 ( 2.5,4 ) ( 2.5,4)(0.5,0)8(1,0)0.3 ( 4.65,4)(1,0)0 ( 4.1,4.3)@xmath152 ( 1.15,4 ) ( 6.5,4 ) ( 6.5,4)(1,0)2 ( 8,4.3)@xmath153 ( 8,4)(1,0)0 ( 4.5,4 ) ( -2,0)(-2.01,0.35)(-1.88,0.68 ) ( -1.88,0.68)(-1.78,1.03)(-1.53,1.29 ) ( -1.53,1.29)(-1.3,1.55)(-1,1.73 ) ( -1,1.73)(-0.69,1.9)(-0.35,1.97 ) ( -0.35,1.97)(0,2.03)(0.35,1.97 ) ( 2,0)(2.01,0.35)(1.88,0.68 ) ( 1.88,0.68)(1.78,1.03)(1.53,1.29 ) ( 1.53,1.29)(1.3,1.55)(1,1.73 ) ( 1,1.73)(0.69,1.9)(0.35,1.97 ) ( 4.8,6)(1,0)0 ( 4.5,4 ) ( -2,0)(-2.36,0.417)(-1.88,0.68 ) ( -1.88,0.68)(-1.47,0.85)(-1.53,1.29 ) ( -1.53,1.29)(-1.54,1.84)(-1,1.73 ) ( -1,1.73)(-0.58,1.6)(-0.35,1.97 ) ( -0.35,1.97)(0,2.35)(0.35,1.97 ) ( 2,0)(2.36,0.417)(1.88,0.68 ) ( 1.88,0.68)(1.47,0.85)(1.53,1.29 ) ( 1.53,1.29)(1.54,1.84)(1,1.73 ) ( 1,1.73)(0.58,1.6)(0.35,1.97 ) ( 4.4,6.5)@xmath31 ( 0.4,2.5)@xmath96 ( 0.8,1.4)(1,0)3 ( 0.8,1.4)(4,-1)3 ( 0.8,1.4 ) ( 2.5,1.41)(-1,0)0 ( 2.75,0.9)(4,-1)0 ( 4,1.4)@xmath46 ( 4,0.5)@xmath47 ( -0.5,-0.5)(a ) contribution to @xmath103 ( 10,7 ) ( 0.5,4)(1,0)2 ( 1.5,4)(1,0)0 ( 0.5,4.3)@xmath151 ( 2.5,4 ) ( 2.5,4)(0.5,0)8(1,0)0.3 ( 4.65,4)(1,0)0 ( 4.1,4.3)@xmath152 ( 7,4.3)@xmath153 ( 8,4 ) ( 6.5,4 ) ( 6.5,4)(1,0)3 ( 9,4)(1,0)0 ( 8.5,4.3)@xmath153 ( 7.4,4)(1,0)0 ( 4.5,4 ) ( -2,0)(-2.01,0.35)(-1.88,0.68 ) ( -1.88,0.68)(-1.78,1.03)(-1.53,1.29 ) ( -1.53,1.29)(-1.3,1.55)(-1,1.73 ) ( -1,1.73)(-0.69,1.9)(-0.35,1.97 ) ( -0.35,1.97)(0,2.03)(0.35,1.97 ) ( 2,0)(2.01,0.35)(1.88,0.68 ) ( 1.88,0.68)(1.78,1.03)(1.53,1.29 ) ( 1.53,1.29)(1.3,1.55)(1,1.73 ) ( 1,1.73)(0.69,1.9)(0.35,1.97 ) ( 4.8,6)(1,0)0 ( 4.5,4 ) ( -2,0)(-2.36,0.417)(-1.88,0.68 ) ( -1.88,0.68)(-1.47,0.85)(-1.53,1.29 ) ( -1.53,1.29)(-1.54,1.84)(-1,1.73 ) ( -1,1.73)(-0.58,1.6)(-0.35,1.97 ) ( -0.35,1.97)(0,2.35)(0.35,1.97 ) ( 2,0)(2.36,0.417)(1.88,0.68 ) ( 1.88,0.68)(1.47,0.85)(1.53,1.29 ) ( 1.53,1.29)(1.54,1.84)(1,1.73 ) ( 1,1.73)(0.58,1.6)(0.35,1.97 ) ( 4.4,6.5)@xmath31 ( 6.3,4 ) ( 7.5,2)@xmath154 ( 6.5,1.3)(-4,-1)3 ( 6.5,1.3)(4,-1)3 ( 6.5,1.3 ) ( 5,0.93)(4,1)0 ( 8.5,0.8)(4,-1)0 ( 2.8,0.5)@xmath47 ( 9.7,0.5)@xmath47 ( 2,-0.5)(b ) contribution to @xmath155 + ( 16,7 ) ( 1,4)(1,0)3.5 ( 6.5,4)(1,0)9.5 ( 3,4 ) ( 4.5,4 ) ( 6.5,4 ) ( 13,4 ) ( 8.5,4 ) ( 3,4)(0,-0.5)5 ( 0,0)(0.125,-0.125)(0,-0.25 ) ( 0,-0.25)(-0.125,-0.375)(0,-0.5 ) ( 8.5,4)(0,-0.5)3 ( 0,0)(0.125,-0.125)(0,-0.25 ) ( 0,-0.25)(-0.125,-0.375)(0,-0.5 ) ( 13,4)(0,-0.5)5 ( 0,0)(0.125,-0.125)(0,-0.25 ) ( 0,-0.25)(-0.125,-0.375)(0,-0.5 ) ( 3,1.5)(1,0)2 ( 3,1.5)(4,-1)2 ( 3,1.5 ) ( 8.5,2.6)(-4,-1)2 ( 8.5,2.6)(4,-1)2 ( 8.5,2.6 ) ( 13,1.5)(-4,-1)2 ( 13,1.5)(4,-1)2 ( 13,1.5 ) ( 1,3.5)@xmath150 ( 15.7,3.5)@xmath156 ( 3.3,2.5)@xmath96 ( 13.3,2.5)@xmath96 ( 8.8,3)@xmath154 ( 5.1,1.4)@xmath46 ( 5.1,0.8)@xmath47 ( 6.2,2.1)@xmath47 ( 10.6,2.1)@xmath47 ( 10.7,0.8)@xmath44 ( 15.1,0.8)@xmath49 ( 3.5,4.3)@xmath151 ( 7.2,4.3)@xmath153 ( 10.5,4.3)@xmath153 ( 5.5,5.3)@xmath31 ( 5.4,4.2)@xmath152 ( 1.8,4)(1,0)0 ( 4,4)(1,0)0 ( 5.7,4)(1,0)0 ( 7.5,4)(1,0)0 ( 11,4)(1,0)0 ( 15.5,4)(1,0)0 ( 4.2,1.5)(-1,0)0 ( 4.36,1.15)(4,-1)0 ( 7.5,2.35)(4,1)0 ( 9.75,2.3)(4,-1)0 ( 12,1.25)(4,1)0 ( 14.5,1.13)(4,-1)0 ( 5.5,4 ) ( -1,0)(-1.005,0.175)(-0.945,0.34 ) ( -0.945,0.34)(-0.89,0.5015)(-0.765,0.645 ) ( -0.765,0.645)(-0.65,0.775)(-0.5,0.865 ) ( -0.5,0.865)(-0.345,0.95)(-0.175,0.985 ) ( -0.175,0.985)(0,1.015)(0.175,0.985 ) ( 1,0)(1.005,0.175)(0.945,0.34 ) ( 0.945,0.34)(0.89,0.5015)(0.765,0.645 ) ( 0.765,0.645)(0.65,0.775)(0.5,0.865 ) ( 0.5,0.865)(0.345,0.95)(0.175,0.985 ) ( 5.7,5)(1,0)0 ( 5.5,4 ) ( -1,0)(-1.18,0.2085)(-0.94,0.34 ) ( -0.94,0.34)(-0.735,0.425)(-0.765,0.645 ) ( -0.765,0.645)(-0.77,0.92)(-0.5,0.865 ) ( -0.5,0.865)(-0.29,0.8)(-0.175,0.985 ) ( -0.175,0.985)(0,1.175)(0.175,0.985 ) ( 1,0)(1.18,0.2085)(0.94,0.34 ) ( 0.94,0.34)(0.735,0.425)(0.765,0.645 ) ( 0.765,0.645)(0.77,0.92)(0.5,0.865 ) ( 0.5,0.865)(0.29,0.8)(0.175,0.985 ) ( 4.5,4)(0.25,0)8(1,0)0.15 ( 3,0)(c ) the whole of the diagram for the process @xmath157 . if we calculate the diagrams shown in figs .
[ fig : wholediagram]-(a ) for @xmath103 and [ fig : wholediagram]-(b ) for @xmath158 , then it means that we count twice the diagram of fig . [
fig : wholediagram]-(c ) into the calculation of the process @xmath159 . in order to avoid doubly counting
, we have to get rid of one of them .
for example , we should not include the contribution of the diagram of fig .
[ fig : wholediagram]-(b ) in @xmath158 .
a similar situation occurs among @xmath160 and @xmath155 .
the penguin contribution to @xmath160 is essentially given by the complex conjugate of that to the corresponding @xmath161 .
however , we must eliminate the contribution from diagrams similar to fig . [
fig : wholediagram]-(a ) from @xmath162 .
finally , we should notice another thing that is important for the double counting problem ; which stage ( _ source _ or _ matter _ ) does the contribution of fig .
[ fig : wholediagram]-(c ) belong to ?
for example , instead of removing the contribution of fig .
[ fig : wholediagram]-(b ) in @xmath158 , we can eliminate the contribution of fig .
[ fig : wholediagram]-(a ) in @xmath163 . for this ambiguity
, we adopt the way to divide the diagram into each stage so that epsilon parameters in each stage will disappear in the limit where @xmath164 after using the off - shell prescription .
the @xmath13 symmetry is recovered and then @xmath165 should disappear in the large susy scale limit .
this is shown analytically .
note that each diagram gives rather large contribution and stays almost constant in the large susy scale limit .
the cancellation among the diagrams is highly nontrivial .
therefore , by checking the cancellation among the diagrams , we can be confident about the legitimacy of our treatment for the internal neutrino lines .
a numerical study to evaluate the epsilon parameters is necessary in order to make it clear whether our naive estimation eq . is correct .
we here assume the universal soft susy breaking at @xmath23 and adopt the scenario of the radiative electroweak symmetry breaking@xcite .
the detail of calculations is shown in appendices .
we scan the values for the soft susy breaking parameters @xmath24 and @xmath166 at the range of 100 - 1,000 gev , and also scan the elements of the neutrino yukawa matrix .
we here take the normal hierarchy for the neutrino mass matrix .
the scatter plots for size of @xmath103 , @xmath155 and @xmath162 are presented in fig .
[ fig : scatter - plots ] .
( 5,5.5 ) , @xmath155 and @xmath162 .
dots are for @xmath167 and triangles are for @xmath168 .
we here fix the soft susy breaking parameters as @xmath169 and @xmath170.,title="fig:",width=188 ] ( -4,4.2)@xmath103 ( -5.5,2.8)@xmath171 ( -3.3,-0.3)br@xmath172 ( 5,5.5 ) , @xmath155 and @xmath162 .
dots are for @xmath167 and triangles are for @xmath168 .
we here fix the soft susy breaking parameters as @xmath169 and @xmath170.,title="fig:",width=188 ] ( -4,4.2)@xmath155 ( -3.3,-0.3)br@xmath172 ( 5,5.5 ) , @xmath155 and @xmath162 .
dots are for @xmath167 and triangles are for @xmath168 .
we here fix the soft susy breaking parameters as @xmath169 and @xmath170.,title="fig:",width=188 ] ( -4,4.2)@xmath162 ( -3.3,-0.3)br@xmath172 all of them take similar behavior . as expected from eq .
, the nlfv parameters correlate with the branching ratio of the process @xmath173 .
however , the size of the parameters is much smaller than that of the naive estimation .
this fact can be understood by the cancellation among the diagrams which contribute to the nlfv interaction . in the @xmath13
symmetric limit , the diagrams for the penguin contribution to the nlfv interaction must cancel each other out because of the gauge symmetry . since the diagrams for nlfv interactions are induced at the @xmath93 scale , in the limit where @xmath174 , the @xmath13 symmetry is assumed to be recovered . even in the case where @xmath175 gev , the cancellation is rather significant .
thus , our naive estimation must be modified so that the additional suppression factor @xmath176 is introduced .
it may be worth pointing out that the epsilon parameters do not strongly depend on the value of @xmath177 unlike the branching ratio of the clfv process .
this fact arises from the difference in the structure of the chirality in each process .
the process @xmath178 is dominated by the diagrams including the left - right mixing of the slepton which is proportional to @xmath177 .
the nlfv processes do not pick up such the left - right mixing term because a chirality flip is not necessary . therefore
, nlfv search may be advantageous in the case where @xmath177 is relatively small .
we also note that it is obvious from eq .
that both nlfv and clfv are enhanced when @xmath25 takes a large value .
we summarize our study and give some discussions .
it is known that magnitude of nlfv couplings can become large enough to be detected at future lbl experiments within model - independent approach .
we here considered the nlfv interactions in the mssmrn under the universal soft susy breaking scenario which is one of the most promising candidates for the physics beyond the sm .
we find that in this scenario the nlfv couplings can not be significant , and hence it is quite difficult to observe these effects at future oscillation experiments .
the reason why they are strongly suppressed is that the @xmath13 gauge symmetry is approximately maintained .
all the particles and the interactions which can generate nlfv interactions respect the @xmath13 symmetry in the limit @xmath179 .
although each diagram to contribute to nlfv interactions can become large , a brilliant cancellation among those diagrams occurs .
therefore , the penguin contributions are strongly suppressed .
we adopt the approximation for the calculation of penguin diagrams which is explained in sec . [
sec : analytic - calc ] . in order to confirm the validity of this approximation
, we need to make the calculation for the process shown in fig .
[ fig : wholediagram]-(c ) by using the method of the field theory
. however , the calculation which we adopt here must be reliable in the sense of the field theory because the consistencies , i.e. , the recovery of @xmath13 gauge symmetry , are obviously maintained in our calculation .
finally , we mention our future work .
since the decay process of a muon differs from that of a pion , we might expect that sizable new physics effect exists only in the decay of pion .
thus , it is necessary to investigate nlfv effects in the mssmrn at superbeam experiments individually .
furthermore , because within the model independent approach , epsilon parameters can be still significant and there are lots of the other models than the constrained mssmrn which can explain the neutrino masses and the lepton mixings , we need to examine such possibilities . _ acknowledgments _
+ j. s. thanks j. w. f. valle for useful discussion .
the work of t. o. is supported in part by japan society for promotion of science no .
. j. s. was supported by the grant - in - aid for scientific research on priority area no.16038202 and 14740168 .
we basically follow the notation which is adopted in ref . @xcite . however , we pay attention to the fact that the mixing matrices to diagonalize the mass matrices for sferimons , chargino and neutralino are complex matrices , in general .
the mass matrices for charged slepton , down - type squark and up - type squark are @xmath180 where @xmath181 here , the indices @xmath4 and @xmath5 are for interaction eigenstates for thier superpartner fermion fields .
we take the basis where the mass matrix for the charged lepton field is diagonal , so that the index for chraged lepton indicates its interaction eigenstate and its mass eigenstate , simultaneously .
the unitary matrix @xmath182 is defined as @xmath183 the relations between the mass eigenstates and the interaction eigenstates are @xmath184 @xmath185 the sneutrino mass term is also given as @xmath186 where @xmath187 where @xmath188 is the neutrino mass matrix which is induced by the seesaw mechanism ; @xmath189 the unitary matrix @xmath190 is defined as @xmath191 the relations between the mass eigenstates and the interaction eigenstates are @xmath192 the chargino mass term in the 2-spinor representation is @xmath193 the diagonalization is done by unitary matrices @xmath194 and @xmath195 as @xmath196 the relations between the interaction eigenstates and the mass eigenstates are @xmath197 where @xmath198 ( @xmath199 ) denotes index for the lorentz spinor for @xmath200 ( @xmath201 ) under @xmath202 .
we here adopt the same rule as that in ref .
the 4-spinors for mass eigenstates are constructed as @xmath203 and then those for interaction eigenstates are @xmath204 the neutralino mass term in 2-spinor representation is @xmath205 the unitary matrix @xmath206 is defined as @xmath207 the relations between the interaction eigenstates and the mass eigenstates are @xmath208 using this 2-spinor @xmath209 , the 4-majorana spinor can be constructed as @xmath210 the interaction eigenstates are @xmath211 the lagrangian for gaugino - sfermion - fermion interactions is described as @xmath212 where the coefficients are @xmath213 , \\ ( n^{(l)}_{l})_{\alpha}^{ax}= & -\frac{g_{2}}{\sqrt{2 } } \left\ { \frac{m_{l_{\alpha}}}{m_{w } \cos\beta } { ( u_{n}^{*})^{a}}_{3 } { ( u_{\tilde{l}}^{*})^{x}}_{\alpha } + 2 { ( u_{n}^{*})^{a}}_{1 } \tan\theta_{w } { ( u_{\tilde{l}}^{*})^{x}}_{\alpha+3 } \right\},\\ ( n^{(\nu)}_{r})_{\alpha a}^{x } = & -\frac{g_{2}}{\sqrt{2 } } \left\ { { ( u_{n})_{a}}^{2 } - { ( u_{n})_{a}}^{1 } \tan\theta_{w } \right\ } { ( u_{\tilde{\nu}}^{*})^{x}}_{\alpha } , \\ ( c^{(d)}_{r})_{\alpha}^{ax } = & g_{2 } \left\ { - { ( u_{r}^{*})^{a}}_{1 } { ( u_{\tilde{u}}^{*})^{x}}_{\alpha } + \frac{m_{u_{\alpha}}}{\sqrt{2 } m_{w } \sin \beta } { ( u_{r}^{*})^{a}}_{2 } { ( u_{\tilde{u}}^{*})^{x}}_{\alpha+3 } \right\},\\ ( c^{(d)}_{l})_{\alpha}^{ax } = & g_{2 } \frac{m_{d_{\alpha}}}{\sqrt{2 } m_{w } \cos \beta } { ( u_{l}^{*})^{a}}_{2 } { ( u_{\tilde{u}}^{*})^{x}}_{\alpha } , \\ ( c^{(u)}_{r})_{\alpha a}^{x } = & g_{2 } \left\ { - { ( u_{l})_{a}}^{1 } { ( u_{\tilde{d}}^{*})^{x}}_{\alpha } + \frac{m_{d_{\alpha}}}{\sqrt{2 } m_{w } \cos \beta } { ( u_{l})_{a}}^{2 } { ( u_{\tilde{d}}^{*})^{x}}_{\alpha+3 } \right\ } , \\ ( c^{(u)}_{l})_{\alpha a}^{x } = & g_{2 } \frac{m_{u_{\alpha}}}{\sqrt{2 } m_{w } \sin \beta } { ( u_{r})_{a}}^{2 } { ( u_{\tilde{d}}^{*})^{x}}_{\alpha } , \\ ( n^{(d)}_{r})_{\alpha a}^{x } = & -\frac{g_{2}}{\sqrt{2 } } \left [ \left\ { - { ( u_{n})_{a}}^{2 } + \frac{1}{3 } { ( u_{n})_{a}}^{1 } \tan\theta_{w } \right\ } { ( u_{\tilde{d}}^{*})^{x}}_{\alpha } + \frac{m_{d_{\alpha}}}{m_{w } \cos\beta } { ( u_{n})_{a}}^{3 } { ( u_{\tilde{d}}^{*})^{x}}_{\alpha+3 } \right ] , \\ ( n^{(d)}_{l})_{\alpha}^{ax } = & -\frac{g_{2}}{\sqrt{2 } } \left\ { \frac{m_{d_{\alpha}}}{m_{w } \cos\beta } { ( u_{n}^{*})^{a}}_{3 } { ( u_{\tilde{d}}^{*})^{x}}_{\alpha } + \frac{2}{3 } { ( u_{n}^{*})^{a}}_{1 } \tan\theta_{w } { ( u_{\tilde{d}}^{*})^{x}}_{\alpha+3 } \right\ } , \\ ( n^{(u)}_{r})_{\alpha a}^{x } = & -\frac{g_{2}}{\sqrt{2 } } \left [ \left\ { { ( u_{n})_{a}}^{2 } + \frac{1}{3}{(u_{n})_{a}}^{1 } \tan\theta_{w } \right\ } { ( u_{\tilde{u}}^{*})^{x}}_{\alpha } + \frac{m_{u_{\alpha}}}{m_{w } \sin\beta } { ( u_{n})_{a}}^{4 } { ( u_{\tilde{u}}^{*})^{x}}_{\alpha+3 } \right ] , \\ ( n^{(u)}_{l})_{\alpha}^{ax } = & -\frac{g_{2}}{\sqrt{2 } } \left\ { \frac{m_{u_{\alpha}}}{m_{w } \sin\beta } { ( u_{n}^{*})^{a}}_{4 } { ( u_{\tilde{u}}^{*})^{x}}_{\alpha } - \frac{4}{3 } { ( u_{n}^{*})^{a}}_{1 } \tan\theta_{w } { ( u_{\tilde{u}}^{*})^{x}}_{\alpha+3 } \right\}.\end{aligned}\ ] ]
we here show the explicit form of the nlfv parameters @xmath103 , @xmath155 and @xmath162 in the mssmrn .
they are calculated from 1-loop diagrams .
the effective coupling comprises two kinds of contribution ; the one comes from the penguin - type diagram associated with @xmath96 boson and the other is the box - type diagram : @xmath214 the penguin - part is represented as @xmath215 and each contribution which is shown in fig . [ fig : penguin - epsilons ] is calculated to be @xmath216 , \label{eq : source - w - penguin - diagram - iii } \\
\mathcal{a}^{\text{($s$-7)}}_{\beta\alpha } = & - \sqrt{2 } \frac{1}{(4\pi)^{2 } } ( c^{(\nu)}_{r})_{\alpha a}^{x } ( n^{(l)*}_{r})^{\beta b}_{x } \nonumber \\ & \times \left [ \delta_{bb } { ( o^{l*})^{ba } } g \left({m_{\tilde{l}_{x } } , m_{\tilde{\chi}^{+}_{a } } , m_{\tilde{\chi}^{0}_{b } } } \right ) - { ( o^{r*})_{b}}^{a } \frac{1}{2 } f \left({m_{\tilde{l}_{x } } , m_{\tilde{\chi}^{+}_{a } } , m_{\tilde{\chi}^{0}_{b } } } \right ) \right ] , \label{eq : source - w - penguin - diagram - iv}\end{aligned}\ ] ] where the functions @xmath217 , @xmath218 , @xmath43 and @xmath219 are defined as @xmath220 with @xmath221 .
the couplings for the chargino - neutralino-@xmath96-boson interaction , @xmath222 and @xmath223 , are given as@xcite @xmath224 the box - part is represented as @xmath225 and each contribution , which is shown in fig .
[ fig : box - epsilons ] , is calculated to be @xmath226 where @xmath227 and @xmath228 are the functions which are given as @xmath229 since the matter of the earth is the neutral for @xmath230 , there is no contribution to @xmath155 from photon - penguin diagrams . the @xmath154-penguin contribution associated with a proton and that with a neutron cancel each other out .
the contributions which need to be taken into account are the @xmath154-penguin contribution associated with an electron and the box contributions : @xmath231 . the diagrams ( @xmath233-a ) and ( @xmath233-d )
are not counted into @xmath155 , which are included in @xmath103 ( cf .
( @xmath234 - 2 ) and ( @xmath234 - 1 ) in fig .
[ fig : penguin - epsilons ] ) .
the diagrams ( @xmath233-b ) and ( @xmath233-c ) cancel each other out after the off - shell prescription ( eq . )
.,width=604 ] each diagrams is calculated to be @xmath235 , \label{eq : z - penguin - epsilonm - ii}\\ \mathcal{a}^{(m\text{-}4)}_{\alpha\beta } = & - \frac{1}{(4 \pi)^{2 } } \left ( 2 \sin^{2 } \theta_{w } -1 \right ) ( c^{(\nu)}_{r})^{x}_{\beta a } ( c^{(\nu)*}_{r})^{\alpha b}_{x } \nonumber \\ & \quad \times \biggl [ { ( o'^{l})^{a}}_{b } g \left ( m_{\tilde{l}_{x } } , m_{\tilde{\chi}^{+}_{a } } , m_{\tilde{\chi}^{+}_{b } } \right ) - \frac{1}{2 } { ( o'^{r})^{a}}_{b } f \left ( m_{\tilde{l}_{x } } , m_{\tilde{\chi}^{+}_{a } } , m_{\tilde{\chi}^{+}_{b } } \right ) \biggr ] .
\label{eq : z - penguin - epsilonm - iv}\end{aligned}\ ] ] the couplings for the chargino - chargino-@xmath154-boson and neutralino - neutralino-@xmath154-boson are@xcite @xmath236 here , we take into account the procedure to resolve the double counting problem which is explained in sec .
[ sec : analytic - calc ] .
we consider the charged current interaction between neutrino beam and the nucleon in the detector as a detection process .
it consists of the penguin contribution and the box contribution ; @xmath243 the penguin contribution can be represented as the complex conjugate of that for @xmath244 .
however , we must eliminate the diagrams which are already counted in the calculation of @xmath155 .
the detail is shown in sec .
[ sec : analytic - calc ] .
@xmath245 gno collaboration , m. altmann _
et al . _ ,
* b490 * ( 2000 ) 16 .
super - kamiokande collaboration , m. smy _
et al . _ ,
phys . rev .
* d69 * ( 2004 ) 011104 ; s. fukuda _ et al .
* 86 * ( 2001 ) 5651 ; ibid .
* 86 * ( 2001 ) 5656 ; phys . lett .
* b539 * ( 2002 ) 179 .
super - kamiokande collaboration , y. ashie _
et al . _ ,
arxiv : hep - ex/0501064 ; phys .
* 93 * ( 2004 ) 101801 ; y. fukuda _ et al . _ ,
* 81 * ( 1998 ) 1562 ; ibid .
* 82 * ( 1999 ) 2644 ; ibid .
* 82 * ( 1999 ) 5194 .
g. giacomelli and a. margiotta , phys . atom .
* 67 * ( 2004 ) 1139 [ yad . fiz .
* 67 * ( 2004 ) 1165 ] .
soudan 2 collaboration , m. sanchez _ et al .
_ , phys . rev . * d68 * ( 2003 ) 113004 .
kamland collaboration , t. araki _
et al . _ ,
arxiv : hep - ex/0406035 ; k. eguchi _
et al . _ ,
* 90 * ( 2003 ) 021802 .
k2k collaboration , m. h. ahn _
et al . _ ,
* 90 * ( 2003 ) 041801 .
p. minkowski , phys . lett .
* b67 * ( 1977 ) 421 .
+ t. yanagida , in _ proceedings of the workshop on unified theory and baryon number of the universe , _ edited by o. sawada and a. sugamoto , report kek-79 - 18 ( 1979 ) .
+ m. gell - mann , p. ramond and r. slansky , in _
supergravity _ , edited by d. z. freedman and p. van nieuwenhuizen ( north - holland , amsterdam , 1979 ) .
see e.g. , j. hisano and d. nomura , phys .
* d59 * ( 1999 ) 116005 ; w. buchmller , d. delphine and f. vissani , phys . lett .
* b459 * ( 1999 ) 171 ; w. buchmller , d. delphine and l. t. handoko , nucl .
b576 * ( 2000 ) 445 ; j. ellis , m. e. gomez , g. k. leontaris , s. lola and d. v. nanopoulos , eur .
j. * c14 * ( 2000 ) 319 ; s. baek , t. goto , y. okada and k. okumura , phys . rev . * d63 * ( 2001 ) 051701 ; j. sato and k. tobe , phys
* d63*(2001 ) 116010 ; j. sato , k. tobe and t. yanagida , phys .
b498 * ( 2001 ) 189 ; j. a. casas and a. ibarra , nucl . phys .
* b618 * ( 2001 ) 171 ; s. davidson and a. ibarra , jhep * 0109 * ( 2001 ) 013 ; s. lavignac , i. masina and c. a. savoy , phys . lett .
* b520 * ( 2001 ) 269 ; j. hisano and k. tobe , phys . lett .
* b510 * ( 2001 ) 197 ; j. ellis and m. raidal , nucl . phys .
* b643 * ( 2002 ) 229 ; t. blazek and s. f. king , nucl . phys .
* b 662 * ( 2003 ) 359 ; s. pascoli , s. t. petcov and c. e. yaguna , phys .
b564 * ( 2003 ) 241 ; s. pascoli , s. t. petcov and w. rodejohann , phys . rev .
* d68 * ( 2003 ) 093007 ; a. masiero , s. profumo , s. k. vempati and c. e. yaguna , jhep * 0403 * ( 2004 ) 046 ; a. masiero , s. k. vempati and o. vives , new j. phys .
* 6 * ( 2004 ) 202 ; s. t. petcov , s. profumo , y. takanishi and c. e. yaguna , nucl . phys .
* b676 * ( 2004 ) 453 ; i. masina and c. a. savoy , arxiv : hep - ph/0501166 ; s. kanemura , k. matsuda , t. ota , t. shindou , e. takasugi and k. tsumura , arxiv : hep - ph/0501228 .
for example , see technical notes in the webpage of the prism project , + http://psux1.kek.jp/~prism .
+ see also the webpage of neutrino factory and muon storage rings at cern , + http://muonstoragerings.web.cern.ch / muonstoragerings/. neutrino factory and muon collider collaboration , c. albright _ et al .
_ , arxiv : physics/0411123 ; c. albright _ et al .
_ , arxiv : hep - ex/0008064 ; neutrino factory and muon collider collabollation , d. ayres _
et al . _ ,
arxiv : physics/9911009 ; d. harris , _ et al .
_ , econf c010630 ( 2001 ) e1001 [ arxiv : hep - ph/0111030 ] .
y. grossman , phys .
b359 * ( 1995 ) 141 .
m. c. gonzalez - garcia , y. grossman , a. gusso and y. nir , phys . rev .
* d64 * ( 2001 ) 096006 .
a. m. gago , m. m. guzzo , h. nunokawa , w. j. c. teves and r. zukanovich funchal , phys
* d64 * ( 2001 ) 073003 .
p. huber and j. w. f. valle , phys . lett .
* b523 * ( 2001 ) 151 .
t. ota , j. sato and n. yamashita , phys .
* d65 * ( 2002 ) 093015 .
s. davidson , c. pena - garay , n. rius and a. santamaria , jhep * 0303 * ( 2003 ) 011 .
see e.g. , e. roulet , phys . rev .
* d44 * ( 1991 ) 935 ; m. m. guzzo , a. masiero and s. t. petcov , phys . lett .
* b260 * ( 1991 ) 154 ; m. m. guzzo and s. t. petcov , phys . lett .
* b271 * ( 1991 ) 172 ; v. d. barger , r. j. n. phillips and k. whisnant , phys .
rev . * d44 * ( 1991 ) 1629 ; s. deglinnocenti and b. ricci , mod .
a8 * ( 1993 ) 471 ; g. l. fogli and e. lisi , astropart . phys . * 2 * ( 1994 ) 91 ; p. i. krastev and j. n. bahcall , _ proceedings of flavor changing neutral currents : present and future studies ( fcnc 97 ) _ , edited by d. b. cline , ( world scientific , singapore , 1997 ) p.259 [ arxiv : hep - ph/9703267 ] ; e. ma and p. roy , phys .
rev . lett .
* 80 * ( 1998 ) 4637 ; s. bergmann , nucl . phys .
* b515 * ( 1998 ) 363 ; s. bergmann and a. kagan , nucl
b538 * ( 1999 ) 368 ; s. bergmann , m. m. guzzo , p. c. de holanda , p. i. krastev and h. nunokawa , phys . rev .
* d62 * ( 2000 ) 073001 ; a. m. gago , m. m. guzzo , p. c. de holanda , h. nunokawa , o. l. g. peres , v. pleitez and r. zukanovich funchal , phys
* d65 * ( 2002 ) 073012 ; m. guzzo , p. c. de holanda , m. maltoni , h. nunokawa , m. a. tortola and j. w. f. valle , nucl . phys . * b629 * ( 2002 ) 479 ; z. berezhiani , r. s. raghavan , a. rossi , nucl . phys .
* b638 * ( 2002 ) 62 ; a. friedland , c. lunardini and c. pena - garay , phys .
b594 * ( 2004 ) 347 ; m. m. guzzo , p. c. de holanda and o. l. g. peres , phys . lett .
* b591 * ( 2004 ) 1 ; o. g. miranda , m. a. trtola and j. w. f. valle , arxiv : hep - ph/0406280 .
see e.g , p. lipari and m. lusignoli , phys .
* d60 * ( 1999 ) 013003 ; g. l. fogli , e. lisi , a. marrone and g. scioscia , phys . rev .
* d60 * ( 1999 ) 053006 ; n. fornengo , m. maltoni , r. t. bayo and j. w. f. valle , phys . rev .
* d65 * ( 2002 ) 013010 ; a. friedland , c. lunardini and m. maltoni , phys . rev .
* d70 * ( 2004 ) 111301 ; m. c. gonzalez - garcia and m. maltoni , phys
* d70 * ( 2004 ) 033010 .
see e.g. , j. w. f. valle , phys .
b199 * ( 1987 ) 432 ; h. nunokawa , y. z. qian , a. rossi and j. w. f. valle , phys . rev . * d54 * ( 1996 ) 4356 ; h. nunokawa , a. rossi and j. w. f. valle , nucl . phys . * b482 * ( 1996 ) 481 ; s. mansour and t. k. kuo , phys . rev .
* d58 * ( 1998 ) 013012 ; g. l. fogli , e. lisi , a. mirizzi and d. montanino , phys . rev . *
d66 * ( 2002 ) 013009 .
g. c. branco , d. delepine , b. nobre , and j. santiago , nucl .
b657 * ( 2003 ) 355 ; a. datta , r. gandhi , b. mukhopadhyaya and poonam mehta , phys . rev .
* d64 * ( 2001 ) 015011 ; l. m. johnson and d. w. mckay , phys . lett .
* b433 * ( 1998 ) 355 . | the effective interactions which violate a lepton flavor accompanied with neutrinos ( nlfv ) are considered .
such a new physics effect is expected to be measured in future neutrino oscillation experiments with long baseline .
they are induced by radiative correction in the framework of the minimal supersymmetric standard model with right - handed neutrinos .
we numerically evaluate the size of the couplings for nlfv interactions in this framework .
the slepton mixing is not only the origin of the lepton flavor violation in the charged lepton sector ( clfv ) but also that of the nlfv .
we find that the nlfv couplings are strongly correlated with the corresponding clfv process , and they are constrained at @xmath0 times smaller than the standard four - fermi couplings . |
analog to analog ( a2a ) compression of signals has recently gathered interest in information theory @xcite . in a2a compression , a high dimensional analog signal @xmath0
is encoded into a lower dimensional analog signal @xmath1 .
the goal is to design the encoding so as to preserve in @xmath2 all the information about @xmath3 , and to obtain successful decoding for a given distortion measure like mse or error probability . in particular ,
the encoding may be corrupted by noise .
it is worth mentioning that when the alphabet of @xmath4 and @xmath5 is finite , this framework falls into traditional topics of information theory such as lossless and lossy data compression , or joint source - channel coding .
the novelty of a2a compression is to consider @xmath4 and @xmath5 to be real valued and to impose regularity constrains on the encoder , in particular linearity , as motivated by compressed sensing @xcite .
the challenge and practicality of a2a compression is to obtain dimensionality reduction , i.e. , @xmath6 , by exploiting a prior knowledge on the signal .
this may be sparsity as in compressed sensing . for @xmath7-sparse signals , and without any stability or complexity considerations ,
it is not hard to see that the dimensionality reduction can be of order @xmath8 . a measurement rate of order @xmath9 has been shown to be sufficient to obtain stable recovery by solving tractable optimization algorithms like convex programming ( @xmath10 minimization ) .
this remarkable achievement has gathered tremendous amount of attention with a large variety of algorithmic solutions deployed over the past years .
the vast majority of the research has however capitalized on a common sparsity model .
several works have explored connections between information theory and compressed sensing , in particular @xcite , however it is only recently @xcite that a foundation of a2a compression has been developed , shifting the attention to probabilistic signal models beyond the sparsity structure .
it is shown in @xcite that under linear encoding and lipschitz - continuous decoding , the fundamental limit of a2a compression is the rnyi information dimension ( rid ) , a measure whose operational meaning had remained marginal in information theory until @xcite . in the case of a nonsingular mixture distribution , the rid is given by the mass on the continuous part , and for the specific case of sparse mixture distributions , this gives a dimensionality reduction of order @xmath8 .
it is natural to ask whether this improvement on compressed sensing is due to potentially complex or non - robust coding strategies .
@xcite shows that robustness to noise is not a limitation of the framework in @xcite .
two other works @xcite have corroborated the fact that complexity may not be a limitation either . in @xcite spatially - coupled matrices are used for the encoding of the signal , leveraging on the analytical ground of spatially - coupled codes and predictions of @xcite . in particular
, @xcite shows that the rid is achieved using approximate message passing algorithm with block diagonal gaussian measurement matrices measurement matrices .
however , the size of the blocks are increasing as the measurement rate approaches the rid . in @xcite , using a new entropy power inequality ( epi ) for integer - valued random variables that was further developed in @xcite , the polarization technique was used to deterministically construct partial hadamard matrices for encoding discrete signals over the reals .
this provides a way to achieve a measurement rate of @xmath11 for signals with a zero rid along with a stable low complexity recovery algorithm .
the case of mixture distributions was however left open in @xcite .
this paper proposes a new approach to a2a compression by means of a polarization theory over the reals .
the use of polarization techniques for sparse recovery was proposed in @xcite for discrete signals , relying on coding strategies over finite fields . in this paper , it is shown that using the rid , one obtains a natural counter - part over the reals of the entropy polarization phenomenon @xcite . specifically , the entropy ( or source ) polarization phenomenon @xcite shows that transforming an i.i.d .
sequence of discrete random variables using an hadamard matrix polarizes the conditional entropies to the extreme values of 0 and 1 ( deterministic and maximally random distributions ) .
we show in this paper that the rid of an i.i.d .
sequence of mixture random variables also polarizes to the two extreme values @xmath12 and @xmath13 ( discrete and continuous distributions ) . to get to this result
, properties of the rid in vector settings and related information measures are first developed .
it is then shown that the rid polarization is , as opposed to the entropy polarization , obtained with an analytical pattern .
in other words , there is no need to rely on algorithms to compute the set of components which tend to 0 or 1 , as this is given by a known pattern equivalent to the bec channel polarization @xcite .
this is then used to obtain universal a2a compression schemes based on explicit partial hadamard matrices .
the current paper focuses on the encoding strategies and on extracting the rid without specifying the decoding strategy .
numerical simulations provide evidence that efficient message passing algorithms may be used in conjunction to the obtained encoders .
finally , the paper extends the realm of a2a compression to a multi signal settings .
techniques of distributed compressed sensing were introduced in @xcite for specific classes of sparse signal models . we provide here an information theoretic framework for general multi signal a2a compression , as a counter part of the slepian&wolf coding problem in source compression @xcite .
a measurement rate region to extract the rid of correlated signals is obtained and is shown to be tight .
the set of reals , integers and positive integers will be denoted by @xmath14 , @xmath15 and @xmath16 respectively .
@xmath17 will denote the set of strictly positive integers . for @xmath18 , @xmath19=\{1,2,\dots , n\}$ ]
denotes the sequence of integers from @xmath13 to @xmath20 . for a set @xmath21
, the cardinality of the set will be denoted by @xmath22 , thus @xmath23|=n$ ] .
all random variables are denoted by capital letters and their realization by lower case letter ( @xmath4 is a realization of the random variable @xmath24 ) . the expected value and the variance of a random variable @xmath24 are denoted by @xmath25 and @xmath26 . for @xmath27
, @xmath28 is a column vector consisting of the random variables @xmath29 and for @xmath30 , we set @xmath28 equal to null . for a discrete random variable @xmath24 with a distribution @xmath31
, @xmath32 denotes the discrete entropy of @xmath24 . for the continuous case
, @xmath33 denotes the differential entropy of @xmath24 . throughout the paper , we assume that all of discrete and continuous random variables have well - defined discrete entropy and differential entropy respectively . for random elements @xmath24 , @xmath34 and @xmath35 , @xmath36 and @xmath37 denote the mutual information of @xmath24 and @xmath34 and the conditional mutual information of @xmath24 and @xmath34 given @xmath35 .
@xmath38 denotes the mutual information of @xmath24 and @xmath34 given a specific realization @xmath39 .
hence , @xmath40 . for simplicity
, we also assume that all of the random variables ( discrete , continuous or mixture ) have finite second order moments .
all probability distributions are assumed to be nonsingular .
hence , in the general case for a random variable @xmath24 , the distribution of @xmath24 can be decomposed as @xmath41 , where @xmath42 and @xmath43 are the continuous and the discrete part of the distribution and @xmath44 is the weight of the continuous part .
thus , @xmath45 and @xmath46 corresponds to the fully discrete and fully continuous case respectively .
for such a probability distribution , the rnyi information dimension is interchangeably denoted by @xmath47 or @xmath48 and is equal to the weight of the continuous part @xmath49 .
there is another representation for a random variable @xmath24 that we will repeatedly use in the paper .
assume @xmath50 is a continuous random variable with probability distribution @xmath42 and @xmath51 is a discrete random variable with probability distribution @xmath43 and @xmath50 and @xmath51 are independent .
let @xmath52 be a binary valued random variable , independent of @xmath50 and @xmath51 with @xmath53 .
it is easy to see that we can represent @xmath24 as @xmath54 , where @xmath55 . in this case
, the random variable @xmath24 will have the distribution @xmath41 .
also , if @xmath56 is a sequence of such random variables with the corresponding binary random variables @xmath57 , @xmath58 : \theta_i=1\}$ ] is a random set consisting of the position of the continuous components of the signal .
similarly , @xmath59\backslash c_\theta$ ] is defined to be the position of the discrete components . for a matrix @xmath60 of a given dimension @xmath61 and a set @xmath62 $ ]
, @xmath63 is a sub - matrix of dimension @xmath64 consisting of those columns of @xmath60 having index in @xmath21 .
similarly , for a vector of random variables @xmath56 , the vector @xmath65 is a sub - vector of @xmath56 consisting of those random variables having index in @xmath21 . for two matrices @xmath66 and @xmath67 of dimensions @xmath68 and @xmath69 , @xmath70 $ ] denotes the @xmath71 matrix obtained by vertically concatenating @xmath66 and @xmath67 . for an @xmath72 and a @xmath73 ,
@xmath74_q=\frac { \lfloor q x \rfloor } { q}$ ] denotes the uniform quantization of @xmath4 by interspacing @xmath75 .
similarly , for a vector of random variables @xmath56 , @xmath76_q$ ] will denote the component - wise uniform quantization of @xmath56 . for @xmath77 and @xmath78 two functions of @xmath79 , @xmath80 or equivalently @xmath81
will be used for @xmath82 similarly , @xmath83 is equivalent to @xmath84 .
an ensemble of single terminal measurement matrices will be denoted by @xmath85 , where @xmath86 is the labeling sequence and can be any subsequence of @xmath87 .
the dimension of the family will be denoted by @xmath88 , where @xmath89 is the number of measurements taken by @xmath90 .
the asymptotic measurement rate of the ensemble is defined by @xmath91 .
we will also work with an ensemble of multi terminal measurement matrices .
we will focus to the two terminal case and the extension to more than two terminals will be straightforward .
we will denote these two terminals by @xmath92 and the corresponding ensemble by @xmath93 with the corresponding dimension @xmath94 and @xmath95 .
the measurement rate vector for this ensemble will be denoted by @xmath96 , where @xmath97 .
let @xmath24 be a random variable with a probability distribution @xmath31 over @xmath14 .
the upper and the lower rid of this random variable are defined as follows : @xmath98_q)}{\log_2(q)},\\ \underline{d}(x)&=\liminf_{q\to \infty } \frac{h([x]_q)}{\log_2(q)}.\end{aligned}\ ] ] by lebesque decomposition or jordan decomposition theorem , any probability distribution over @xmath14 like @xmath31 can be written as a convex combination of a discrete part , a continuous part and a singular part , namely , @xmath99 where @xmath43 , @xmath42 and @xmath100 denote the discrete , continuous and the singular part of the distribution and @xmath101 and @xmath102 . in @xcite
, rnyi showed that if @xmath103 , namely , there is no singular part in the distribution and @xmath104 for some @xmath105 $ ] , then the rid is well - defined and @xmath106 .
moreover , he proved that if @xmath56 is a continuous random vector then @xmath107_q)}{\log_2(q)}=n$ ] , implying the rid of @xmath20 for the @xmath20-dimensional continuous random vector .
our objective is to extend the definition of rid for arbitrary vector random variables , which are not necessarily continuous .
to do so , we first restrict ourselves to a rich space of random variables with well - defined rid . over this space
, it will be possible to give a full characterization of the rid as we will see in a moment .
let @xmath108 be a standard probability space .
the space @xmath109 is defined as @xmath110 , where @xmath111 is the set of all nonsingular random variables and for @xmath112 , @xmath113 is the space of @xmath20-dimensional random vectors defined as @xmath114 it is not difficult to see that all @xmath20-dimensional vector random variables , singular or nonsingular , can be well approximated in the space @xmath115 , for example in @xmath116-sense . however ,
this is not sufficient to fully characterize the rid .
specially , the rid is discontinuous in @xmath117 topology , @xmath118 .
for example , we can construct a sequence of fully discrete random variables in @xmath115 converging to a fully continuous random variable in @xmath119 , whereas the rid of the sequence is @xmath12 and does not converge to @xmath13 .
although we have such a mathematical difficulty in giving a characterization of the rid , we think that the space @xmath115 is rich enough for modeling most of the cases that we encounter in applications . over @xmath115
, we will generalize the definition of the rid to include joint rid , conditional rid and rnyi information defined as follows .
let @xmath56 be a random vector in @xmath115 .
the joint rid of @xmath56 provided that it exists , is defined as @xmath120_q)}{\log_2(q)}.\ ] ] let @xmath121 be a random vector in @xmath115 .
the conditional rid of @xmath56 given @xmath122 and rnyi information of @xmath123 about @xmath56 , provided they exist , are defined as follows : @xmath124_q|y_1^m)}{\log_2(q)}\\ & i_r(x_1^n;y_1^n)=d(x_1^n)-d(x_1^n|y_1^m).\end{aligned}\ ] ] generally , it is difficult to give a characterization of rid for a general multi - dimensional distribution because it can contain probability mass over complicated subsets or sub - manifolds of lower dimension .
however , we will show that the vector rnyi information dimension is well - defined for the space @xmath115 . in order to give the characterization of rid over @xmath115
, we also need to define some concepts from linear algebra of matrices , namely , for two matrices of appropriate dimensions , we propose the following definition of the influence " of one matrix on another matrix and residual " of one matrix given another matrix .
let @xmath66 and @xmath67 be two arbitrary matrices of dimension @xmath68 and @xmath69 .
also let @xmath125 $ ] .
the influence of the matrix @xmath67 on the matrix @xmath66 and the residual of the matrix @xmath66 given @xmath67 over the column set @xmath126 are defined to be @xmath127&={\mathrm{rank}}([a;b]_k)-{\mathrm{rank}}(a_k),\\ r(a;b)[k]&={\mathrm{rank}}([a;b]_k)-{\mathrm{rank}}(b_k).\end{aligned}\ ] ] it is easy to check that @xmath128 $ ] is the amount of increase of the rank of the matrix @xmath129 by adding rows of the matrix @xmath130 and @xmath131 $ ] is the residual rank of the matrix @xmath129 knowing the rows of the matrix @xmath130 .
moreover , one can easily check that @xmath128=r(b;a)[k]$ ] .
[ rid_maintheorem ] let @xmath121 be a random vectors in the space @xmath115 , namely , there are i.i.d .
nonsingular random variables @xmath132 and two matrices @xmath66 and @xmath67 of dimension @xmath133 and @xmath134 such that @xmath135 and @xmath136 .
let @xmath137 be the representation for @xmath138 $ ] .
then , we have 1 .
@xmath139 , 2 .
@xmath140\}$ ] , where @xmath141 : \theta_i=1\}$ ] is the random set consisting of the position of continuous components .
notice that the results intuitively make sense , namely , for a specific realization @xmath142 if @xmath143 we can neglect @xmath144 because it is fully discrete and does not affect the rid .
moreover , over the continuous components the resulting contribution to the rid is equal to the rank of the matrix @xmath145 , which is the effective dimension of the space over which the continuous random variable @xmath146 is distributed . finally ,
all of these contributions are averaged over all possible realizations of @xmath147 . using theorem [ rid_maintheorem ] , it is possible to prove a list of properties of the rid .
[ rid_extensions ] let @xmath121 be a random vector in @xmath115 as in theorem [ rid_maintheorem ] .
then , we have the following properties : 1 .
@xmath148 for any arbitrary invertible matrix @xmath149 of dimension @xmath150 .
2 . @xmath151 .
3 . @xmath152 .
4 . @xmath153 and @xmath154 if and only if @xmath56 and @xmath122 are independent after removing discrete common parts , namely , those @xmath155 $ ] that are fully discrete .
further investigation also shows that we have a very nice duality between the discrete entropy and the rid as depicted in table [ tab : duality1 ] .
as we will see in subsection [ results : single ] and [ results : multi ] , this duality can be generalized to include some of the theorems in classical information theory like single terminal and multi terminal ( slepian & wolf ) source coding problems .
.duality between @xmath156 and @xmath157 [ cols="^,^ " , ]
in this section , we will give a brief overview of the results proved in the paper .
subsection [ results : rid_polarize ] is devoted to the results obtained for the polarization of the rnyi information dimension .
these results are used in subsections [ results : single ] and [ results : multi ] to study _
compression _ problem from an information theoretic point of view .
subsection [ results : single ] considers the single terminal case whereas subsection [ results : multi ] is devoted to the multi terminal case . before stating the polarization result for the rid , we define the @xmath158-dimensional erasure process as follows .
let @xmath159 $ ] .
an `` erasure process '' with initial value @xmath160 is defined as follows .
1 . @xmath161 . @xmath162 and @xmath163 .
2 . let @xmath164 , for some arbitrary @xmath165-valued sequence @xmath166 .
define @xmath167 notice that using the @xmath165 labeling , we can construct a binary tree where each leaf of the tree is assigned a specific @xmath165-valued sequence .
let @xmath168 be a sequence of i.i.d .
uniform @xmath165-valued random variables . by replacing @xmath169 for @xmath165-labeling @xmath166 in the definition of the erasure process
, we obtain a stochastic process @xmath170 .
let @xmath171 be the @xmath172-field generated by @xmath169 . using the bec polarization @xcite
, we have the following results : 1 .
@xmath173 is a positive bounded martingale .
@xmath174 converges to @xmath175 with @xmath176 .
3 . for any @xmath177 , @xmath178 , where @xmath179 is the number of all possible cases that @xmath174 can take . let @xmath18 and @xmath179 .
assume that @xmath180 is a sequence of i.i.d .
nonsingular random variables with a rid equal to @xmath48 and let @xmath181 , where @xmath182 is the hadamard matrix of order @xmath86 . for @xmath183
$ ] , let us define @xmath184 .
assume that @xmath166 is the binary expansion of @xmath185 . by replacing @xmath12 by @xmath186 and @xmath13 by @xmath187
, we can equivalently represent @xmath188 be a sequence of @xmath165 values , namely , @xmath189 .
similar to the erasure process , we can convert @xmath190 to a stochastic process @xmath191 by using i.i.d .
uniform @xmath165-valued random variables @xmath169 .
we have the following theorem . [ single_rid_polarization ]
@xmath192 is an erasure stochastic process with initial value @xmath48 polarizing to @xmath193 . for @xmath18 and @xmath179 ,
let @xmath194 be a sequences of random vectors in the space @xmath115 , with joint and conditional rid @xmath195 , @xmath196 and @xmath197 .
let @xmath181 and assume that @xmath198 .
let us define two processes @xmath190 and @xmath199 as follows .
@xmath200,\\ j_n(i)&=d(w_i|w_1^{i-1},z_1^n ) , i\in [ n].\end{aligned}\ ] ] similarly , we can label @xmath190 and @xmath199 by a sequence of @xmath166 and convert them to stochastic processes @xmath191 and @xmath201 . by this definition
, we have the following theorem . [ multi_rid_polarization ]
@xmath192 and @xmath202 are erasure stochastic processes with initial value @xmath48 and @xmath197 , both polarizing to @xmath193 .
in the @xmath203 terminal case @xmath204 for a @xmath203 terminal source @xmath205 , using a similar method it is possible to construct erasure processes with initial values @xmath206 , polarizing to @xmath193 . in this subsection
, we will use the properties of the rid developed in section [ section : rid ] to study the a2a compression of memoryless sources .
we assume that we have a memoryless source with some given probability distribution .
the idea is to capture the information of the source , to be made clearer in a moment , by taking some linear measurements . as is usual in information theory , we are mostly interested in asymptotic regime for large block lengths . to do so
, we will use an ensemble of measurement matrices to analyze the asymptotic behavior .
we will also define the notion of rep ( restricted iso - entropy property ) for an ensemble of measurement matrices .
this subsection is devoted to the single terminal case .
the results for the multi terminal case will be given in subsection [ results : multi ] .
we are mostly interested to the the measurement rate region of the problem in order to successfully capture the source .
let @xmath180 be a sequence of i.i.d .
random variables with a probability distribution @xmath31 ( discrete , mixture or continuous ) over @xmath14 , and let @xmath207_q$ ] for @xmath208 .
the family of measurement matrices @xmath209 , indexed with a subsequence of @xmath87 and with dimension @xmath88 , is @xmath210-rep(@xmath31 ) with the measurement rate @xmath211 if @xmath212 to give some intuitive justification for the rep definition , let us assume that all of the measurements are captured with a device with finite precision @xmath213 for some @xmath214 . in that case , although the potential information of the signal , in terms of bits , can be very large , but what we effectively observe through the finite precision device is only @xmath215_{q_0})$ ] .
in such a setting , the ratio of the information we lose after taking the measurements , assuming that some genie gives us the infinite precision measurement captured from the signal , is exactly what we have in the definition of rep , namely , @xmath216 where we assume that @xmath207_{q_0}$ ] .
this might be a reasonable model for application because pretty much this is what happens in reality .
the problem with this model is that it is not invariant under some obvious transformations like scaling .
for example , assume that we are scaling the signal by some real number . in this case , through some simple examples it is possible to show that the ratio in ( [ information_ratio ] ) can change considerably .
there are two approaches to cope with this problem .
one is to scale the signal with a desired factor to match it to the finite precision quantizer , which in its own can be very interesting to analyze but probably will be two complicated .
the other way , is to take our approach and develop a theory for the case in which the resolution is high enough so that the quality measure proposed in ( [ information_ratio ] ) is not affected by the shape of the distribution of the signal .
notice that in the fully discrete case , the rep definition is simplified to the equivalent form @xmath217 for a non discrete source with strictly positive rid , @xmath218 , if we divide the numerator and the denumerator in the expression ( [ rep_definition ] ) by @xmath219 , take the limit as @xmath79 tends to infinity and use the definition of the rid , we get the equivalent form @xmath220 interestingly , this implies that in the high resolution regime that we are considering for analysis , the information isometry ( keeping more than @xmath221 ratio of the information of the signal ) is equivalent to the rnyi isometry .
moreover , from the properties of the rid , it is easy to see that this rep measure meets some of the invariance requirements that we expect .
for example , it is scale invariant and any invertible linear transformation of the input signal @xmath180 keeps the @xmath210-rep measure unchanged . we can also extend the definition when the probability distribution of the source is not known exactly but it is known to belong to a given collection of distributions @xmath222 .
assume @xmath223 is a class of nonsingular probability distributions over @xmath14 .
the family of measurement matrices @xmath209 , indexed with a subsequence of @xmath87 and with dimension @xmath88 , is @xmath210-rep ( @xmath222 ) for measurement rate @xmath211 if it is @xmath210-rep ( @xmath224 ) for every @xmath225 . now that we have the required tools and definitions , we give a characterization of the required measurement rate in order to keep the information isometry .
similar to all theorems in information theory , we do this using the converse " and achievability " parts .
[ mixture_converse ] let @xmath180 be a sequence of i.i.d .
random variables in @xmath115 .
suppose @xmath209 is a family of @xmath210-rep(@xmath31 ) measurement matrices of dimension @xmath88 , then @xmath226 .
this result implies that to capture the information of the signal the asymptotic measurement rate must be approximately greater then the rid of the source .
this in some sense is similar to the single terminal source coding problem in which the encoding rate must be grater then the entropy of the source .
this again the emphasizes the analogy between @xmath156 and @xmath157 .
moreover , in the discrete case , @xmath227 , the result is trivial .
it was proved in @xcite that under linear encoding and block error probability distortion condition , the measurement rate must be higher than the rid of the source , @xmath228 .
theorem [ mixture_converse ] strengthen this result stating that @xmath229 must hold even under the milder @xmath210-rep restriction on the measurement ensemble .
theorem [ mixture_converse ] puts a lower bound on the measurement rate in order to keep the @xmath210-rep property .
however , it might happen that there is no measurement family to achieve this bound . fortunately
, as we will see , it is possible to deterministically truncate the family of hadamard matrices to obtain a measurement family with @xmath210-rep property and measurement rate @xmath48 .
this is summarized in the following two theorems .
notice that in the fully continuous case as theorem [ mixture_converse ] implies , the feasible measurement rate is approximately @xmath13 which for example can be achieved with any complete orthonormal family , thus no explicit construction is necessary . for the noncontinuous case , we will distinguish between the fully discrete case and the mixture case because they need different proof techniques .
theorem [ achievability_discrete ] and [ achievability_hadamard ] summarize the results .
[ achievability_discrete ] let @xmath180 be a sequence of i.i.d .
discrete integer - valued random variables .
then , for any @xmath230 , there is a family of @xmath210-rep(@xmath31 ) partial hadamard matrices of dimension @xmath88 , for @xmath179 with @xmath231 .
[ achievability_hadamard ] let @xmath180 be a sequence of i.i.d .
random variables in @xmath115 .
then , for any @xmath230 , there is a family of @xmath210-rep(@xmath31 ) partial hadamard matrices of dimension @xmath88 , for @xmath179 with @xmath232 .
we have also the general result in theorem [ universal_mixture ] which implies that we can construct a family of truncated hadamard matrices which is @xmath210-rep for a class of distributions .
[ universal_mixture ]
let @xmath222 be a family of probability distributions with strictly positive rid .
then , for any @xmath230 , there is a family of @xmath210-rep ( @xmath222 ) partial hadamard matrices of dimension @xmath88 , for @xmath179 , with @xmath233 .
theorem [ universal_mixture ] implies that there is a fixed ensemble of measurement matrices capable of capturing the information of the all of the distributions in the family @xmath222 .
this is very useful in applications because usually taking the measurements is costly and most of the time we do not have the exact distribution of the signal .
if each distribution needs its own specific measurement matrix , we have to do several rounds of the measurement each time taking the measurements compatible with one specific distribution and do the recovery process for that specific distribution .
the benefit of theorem [ universal_mixture ] is that one measurement ensemble works for all of distributions .
it is also good to notice that although the measurement ensemble is fixed , the recovery ( decoding ) process might need to know the exact distribution of the signal in order to have successful recovery . in this section ,
our goal is to extend the a2a compression theory from the single terminal case to the multi terminal case . in the multi terminal
setting , we have a memoryless source which is distributed in more than one terminal and we are going to take linear measurements from different terminals in order to capture the information of the source .
we are again interested in an asymptotic regime for large block lengths .
to do so , we will use an ensemble of distributed measurement matrices that we will introduce in a moment .
similar to the single terminal case , we are interested in the measurement rate region of the problem , namely , the number of measurements that we need from different terminals in order to capture the signal faithfully . we will analyze the problem for two terminal case .
the extension to more than two terminals is straightforward .
let @xmath234 be a two terminal memoryless source with @xmath235 being in @xmath115 .
the family of distributed measurement matrices @xmath236 , indexed with a subsequence of @xmath87 , is @xmath210-rep @xmath237 for the measurement rate @xmath96 if @xmath238_q,[y_1^n]_q|\phi^x_n x_1^n , \phi^y_n y_1^n)}{h([x_1^n]_q,[y_1^n]_q ) } \leq \epsilon,\\ & \limsup _ { n \to \infty } \frac{m^x_n}{n}\leq\rho_x,\
\ \limsup_{n \to \infty } \frac{m^y_n}{n}\leq \rho_y.\nonumber\end{aligned}\ ] ] if @xmath239 is a random vector in @xmath115 with @xmath240 , similar to what did in the single terminal case , dividing the numerator and the denumerator in the expression ( [ multiterminal_formula ] ) by @xmath219 and taking the limit as @xmath79 tends to infinity , we get the equivalent definition @xmath241 which implies the equivalence of the information isometry and the rnyi isometry .
notice that in the fully discrete case , the definition above is simplified to the equivalent form @xmath242 we can also extend the definition to a class of probability distributions . assume that @xmath223 is a class of nonsingular probability distributions in @xmath115 .
the family of measurement matrices @xmath243 is @xmath210-rep ( @xmath222 ) for measurement rate @xmath96 if it is @xmath210-rep ( @xmath224 ) for every @xmath225 .
let @xmath239 be a two dimensional random vector in @xmath115 with a distribution @xmath244 .
the rnyi information region of @xmath244 is the set of all @xmath245 ^ 2 $ ] satisfying @xmath246 assume that @xmath222 is a class of two dimensional random vectors from @xmath115 .
the rnyi information region of the class @xmath222 is the intersection of the rnyi information regions of the distributions in @xmath222 .
similar to the single terminal case , we are interested in the rate region of the problem .
we have the following converse and achievability results .
[ converse_theorem_multi ] let @xmath234 be a two - terminal memoryless source with @xmath235 being in @xmath115 .
assume that the distributed family of measurement matrices @xmath243 is @xmath210-rep with a measurement rate @xmath96 .
then , @xmath247 this rate region is very similar to the rate region of the distributed source coding ( slepian&wolf ) problem with the only difference that the discrete entropy has been replaced by the rid , which again emphasizes the analogy between the discrete entropy and the rid .
similar to the slepian&wolf problem , we call @xmath248 the dominant face of the measurement rate region .
[ achievability_discrete_multi ] let @xmath234 be a discrete two - terminal memoryless source .
then there is a family of @xmath210-rep partial hadamard matrices @xmath243 with @xmath249 .
[ achievability_hadamard_multi ] let @xmath234 be a two - terminal memoryless source with @xmath235 belonging to @xmath115 .
given any @xmath96 satisfying @xmath250 there is a family of @xmath210-rep partial hadamard matrices with measurement rate @xmath96 .
we have also the general result in theorem [ universal_mixture_multi ] which implies that we can construct a family of truncated hadamard matrices which is @xmath210-rep for a class of distributions . [ universal_mixture_multi ]
let @xmath222 be a family of two dimensional probability distributions in @xmath115 .
then , for any @xmath96 in the measurement region of @xmath222 , there is a family of partial hadamard matrices which is @xmath210-rep ( @xmath222 ) with a measurement rate @xmath96 .
in this section , we will give a brief overview of the techniques used to prove the results .
we will divide this section into three subsections . in subsection
[ prooftech : rid ] , we will overview the proof techniques for the rid .
subsection [ prooftech : single ] and [ prooftech : multi ] will be devoted to proof ideas and intuitions about the a2a compression problem in the single and multi terminal case .
in this section we will prove theorem [ rid_maintheorem ] and [ rid_extensions ] and we will give further intuitions about the rid over the space @xmath115 .
* proof of theorem [ rid_maintheorem ] : * to prove the first part of the theorem , notice that @xmath251_q ) \doteq h([x_1^n]_q , \theta_1^k ) \doteq h([x_1^n]_q|\theta_1^k),\ ] ] because @xmath252 . as @xmath253 and takes finitely many values , it is sufficient to show that for any realization @xmath142 , @xmath254_q|\theta_1^k)}{\log_2(q)}={\mathrm{rank}}(a_{{c_\theta } } ) .
\end{aligned}\ ] ] taking the expectation over @xmath147 , we will get the result . to prove ( [ rid_maintheorem_formula1 ] ) , notice that @xmath255_q|\theta_1^k)&= h([a_{{c_\theta}}u_{{c_\theta}}+ a_{{\bar{c}_\theta}}v_{{\bar{c}_\theta}}]_q)\nonumber \\ & \doteq h([a_{{c_\theta}}u_{{c_\theta}}+ a_{{\bar{c}_\theta}}v_{{\bar{c}_\theta}}]_q|v_{{\bar{c}_\theta}})\label{rid_maintheorem_formula2}\\ & \doteq h([a_{{c_\theta}}u_{{c_\theta}}]_q),\label{rid_maintheorem_formula3}\end{aligned}\ ] ] where we used @xmath256 .
we also used the fact that knowing @xmath257 , @xmath258_q$ ] and @xmath259_q$ ] are equal up to finite uncertainty . specifically , suppose @xmath260 is the minimum number of lattices of size @xmath75 required to cover @xmath261^{|{{\bar{c}_\theta}}|}$ ] , which is a finite number .
then @xmath262_q | v_{{\bar{c}_\theta } } , [ a_{{c_\theta}}u_{{c_\theta}}+ a_{{\bar{c}_\theta}}v_{{\bar{c}_\theta}}]_q ) \leq \log_2(l),\ ] ] which implies ( [ rid_maintheorem_formula2 ] ) and ( [ rid_maintheorem_formula3 ] ) .
generally @xmath145 is not full rank .
assume that the rank of @xmath145 is equal to @xmath158 and let @xmath263 be a subset of linearly independent rows .
it is not difficult to see that knowing @xmath264_q$ ] there is only finite uncertainty in the remaining components of @xmath258_q$ ] , which is negligible compared with @xmath219 as @xmath79 tends to infinity .
therefore , we obtain @xmath255_q|\theta_1^k)&\doteq h([a_{{c_\theta}}u_{{c_\theta}}]_q)\\ & \doteq h([a_m u_{{c_\theta}}]_q)\\ & \doteq m \log_2(q).\end{aligned}\ ] ] thus , taking the limit as @xmath79 tends to infinity , we obtain @xmath265|\theta_1^k)}{\log_2(q)}={\mathrm{rank}}(a_{{c_\theta}}).\ ] ] also , taking the expectation with respect to @xmath147 , we obtain @xmath266 , which is the desired result . to prove the second part of the theorem , notice that @xmath251_q | y_1^m ) \doteq h([x_1^n]_q | y_1^m , \theta_1^k).\ ] ] for a specific realization @xmath142 we have @xmath267_q | y_1^m,\theta_1^k)\\ & = h([a_{{c_\theta}}u_{{c_\theta}}+ a_{{\bar{c}_\theta}}v_{{\bar{c}_\theta}}]_q | b_{{c_\theta}}u_{{c_\theta}}+ b_{{\bar{c}_\theta}}v_{{\bar{c}_\theta}})\\ & \doteq h([a_{{c_\theta}}u_{{c_\theta}}+ a_{{\bar{c}_\theta}}v_{{\bar{c}_\theta}}]_q | b_{{c_\theta}}u_{{c_\theta}}+ b_{{\bar{c}_\theta}}v_{{\bar{c}_\theta } } , v_{{\bar{c}_\theta}})\\ & \doteq h([a_{{c_\theta}}u_{{c_\theta}}]_q | b_{{c_\theta}}u_{{c_\theta}}).\end{aligned}\ ] ] generally , @xmath145 is not full - rank .
let @xmath263 be the set of all linearly independent rows of @xmath145 of size @xmath158 .
then @xmath262_q | b_{{c_\theta}}u_{{c_\theta}})\doteq h([a_m u_{{c_\theta}}]_q|b_{{c_\theta}}u_{{c_\theta}}).\ ] ] it may happen that some of the rows of @xmath263 can be written as a linear combination of rows of @xmath268 . let @xmath269 be the remaining matrix after dropping @xmath270 predictable rows of @xmath263
. given , @xmath271 , @xmath272 has a continuous distribution thus @xmath273_q|b_{{c_\theta}}u_{{c_\theta } } ) \doteq r \log_2(q).\ ] ] it is easy to check that @xmath274 is exactly @xmath275 $ ] .
therefore , taking the expectation with respect to @xmath147 , we get @xmath276\}.\ ] ] we also get the following corollary , which shows the additive property of the rid for the independent random variables from @xmath115 .
let @xmath56 be independent random variables from @xmath115
. then @xmath277 .
notice that we can simply write @xmath278 , where @xmath279 is the identity matrix of order @xmath86 .
therefore , by the rank characterization for the rid , we have @xmath280)\}={\mathbb{e}}\{\sum_{i=1}^n \theta_i \}=\sum_{i=1}^n d(x_i),\end{aligned}\ ] ] where we used the fact that the columns of @xmath279 are linearly independent thus adding a column increases the rank by @xmath13 .
therefore , the rank of @xmath281 is equal to the number of @xmath13 s is @xmath282 , namely , @xmath283 . using the results of theorem [ rid_maintheorem ]
, we can prove theorem [ rid_extensions ] .
* proof of theorem [ rid_extensions ] : * for part @xmath13 , the proof is simple by considering the rank characterization . we know that @xmath135 and @xmath284 .
moreover , @xmath285 thus @xmath286 .
as @xmath149 is invertible @xmath287 , thus we get the result .
for part @xmath288 , notice that for any realization @xmath142 and the corresponding set @xmath289 , @xmath290_{{c_\theta}})&={\mathrm{rank}}(a_{{c_\theta } } ) + r(b;a)[{{c_\theta}}]\\ & = { \mathrm{rank}}(b_{{c_\theta } } ) + r(a;b)[{{c_\theta}}].\end{aligned}\ ] ] taking the expectation over @xmath147 , we get the desired result @xmath291 for part @xmath292 , using the chain rule result from part @xmath288 and applying the definition of @xmath293 , we get @xmath294 which shows the symmetry of @xmath295 with respect to @xmath56 and @xmath122 .
for part @xmath296 , notice that for a specific realization @xmath142 , a simple rank check shows that @xmath275 \leq { \mathrm{rank}}(a_{{c_\theta}})$ ] . taking the expectation over @xmath147 , we get @xmath297 .
if @xmath56 and @xmath122 are independent , the equality follows from the definition .
for the converse part , notice that if @xmath298 is fully discrete then @xmath299 .
similarly , if @xmath122 is fully discrete then @xmath300 and using the identity @xmath301 , we get the equality .
this case is fine because after removing the discrete @xmath155 $ ] , either @xmath56 or @xmath122 is equal to @xmath12 , namely , a deterministic value , and the independence holds . assume that none of @xmath56 or @xmath122 is fully discrete . without loss of generality , let @xmath302 be the non - discrete random variables among @xmath132 and let @xmath303 and @xmath304 be the resulting random vectors after dropping the discrete constituents , namely , we have @xmath305 and @xmath306 , where @xmath269 and @xmath307 are the matrices consisting of the first @xmath274 columns of @xmath66 and @xmath67 respectively .
it is easy to check that @xmath308 and @xmath309 .
thus it remains to show that @xmath303 and @xmath304 are independent .
as we have dropped all of the discrete components , the resulting @xmath310 $ ] are @xmath13 with strictly positive probability .
this implies that for any realization of @xmath311 and the corresponding @xmath312 , @xmath313={\mathrm{rank}}(a_{r,{{c_\theta}}})$ ] .
in particular , this holds for any @xmath312 of size @xmath13 , namely , for any column of @xmath269 and @xmath307 , which implies that if @xmath269 has a non - zero column the corresponding column in @xmath307 must be zero and if @xmath307 has a non - zero column then the corresponding column in @xmath269 must be zero .
this implies that @xmath303 and @xmath304 depend on disjoint subsets of the random variables @xmath302 .
therefore , they must be independent . in this section , we will prove the polarization of the rid in the single and multi terminal case as stated in theorem [ single_rid_polarization ] and theorem [ multi_rid_polarization ] .
the main idea is to use the recursive structure of the hadamard matrices and the rank characterization of the rid in the space @xmath115 . for the initial value
, we have @xmath314 .
let @xmath18 and @xmath179 . to simplify the proof , instead of the hadamard matrices , @xmath156 , we will use shuffled hadamard matrices , @xmath315 , constructed as follows : @xmath316 and @xmath317 is constructed from @xmath318 as follows @xmath319 where @xmath320 , @xmath183 $ ] denotes the @xmath321-th row of the @xmath318 .
let @xmath56 be as in theorem [ single_rid_polarization ] and let @xmath322 , where @xmath182 is replaced by @xmath318 .
also , let @xmath323 , @xmath183 $ ] .
we first prove that @xmath324 is also an erasure process with initial value @xmath325 and evolves as follows @xmath326 where @xmath327 $ ] with the corresponding @xmath165-labeling @xmath166 .
also , let @xmath328 and @xmath329 denote the first @xmath185 and the first @xmath321 rows of @xmath318 .
also , let @xmath320 denote the @xmath321-th row of @xmath318 .
thus , we have @xmath330 and @xmath331 . as @xmath180 are i.i.d
. nonsingular random variables , it results that @xmath332 belong to the space @xmath115 generated by the @xmath180 random variables .
notice that using the rank characterization for the rid over @xmath115 , we have @xmath333\},\end{aligned}\ ] ] where @xmath334 \in \{0,1\}$ ] is the amount of increase of rank of @xmath335 by adding @xmath320 .
now , consider the stage @xmath336 , where we have the shuffled hadamard matrix @xmath317 .
consider the row @xmath337 which corresponds to the row @xmath338 of @xmath317 . now ,
if we look at the first block of the new matrix , we simply notice that adding @xmath320 has the same effect in increasing the rank of this block as it had in @xmath318 .
a similar argument holds for the second block . moreover , adding @xmath320 increases the rank of the matrix if it increases the rank of either the first or the second block or both .
let @xmath339 denote the random rank increase in @xmath328 by adding @xmath320 , then we have @xmath340 @xmath282 and @xmath341 are i.i.d .
random variables and a simple check shows that @xmath342 and @xmath343 are also i.i.d .. taking the expectation value , we obtain @xmath344 moreover , if we denote @xmath345 , then by the structure of @xmath318 it is easy to see that @xmath346 and @xmath347 can be written as follows : @xmath348 using the chain rule for the rid , we have @xmath349 which along with ( [ plus_part ] ) , implies that @xmath350 therefore , @xmath324 evolves like an erasure process with initial value @xmath48 .
now , notice that the only difference between @xmath182 and @xmath318 is the permutation of the rows , namely , there is a row shuffling matrix @xmath351 such that @xmath352 .
it was proved in @xcite that @xmath351 and @xmath182 commute , which implies that @xmath353 .
however , notice that @xmath180 is an i.i.d .
sequence and @xmath354 is again an i.i.d .
sequence with the same distribution as @xmath180 .
in particular , adding or removing @xmath351 does not change the rid values , which implies that for @xmath181 and @xmath184 , @xmath355 . therefore , @xmath356 is also be an erasure process with initial value @xmath48 , which polarizes to @xmath193 . using a similar technique
, we can prove theorem [ multi_rid_polarization ] .
the main idea is that @xmath239 are correlated random variables in the space @xmath115 and they can be written as a linear combination of i.i.d .
nonsingular random variables . for the initial value
, we have @xmath314 and @xmath357 . as @xmath234 is a memoryless source , similar to the single terminal case , it is easy to see that @xmath356 is an erasure process with initial value @xmath325 and it remains to show that @xmath358 is also an erasure process but with initial value @xmath359
. let @xmath328 , @xmath329 and @xmath320 denote the first @xmath185 rows , the first @xmath321 rows , and the @xmath321-th row @xmath318 .
as @xmath360 there is a sequence of i.i.d .
random variables @xmath361 and two vectors @xmath362 and @xmath363 such that @xmath364 and @xmath365 .
as @xmath234 is memoryless , there is a concatenation of sequence of i.i.d .
copies of @xmath361 , @xmath366 , such that @xmath367 e,\\ w_{1}^{n}&=\tilde{h}_n y_1^n=[(b_n h_n ) \otimes ( b_1^k)^t ] e,\end{aligned}\ ] ] where @xmath368 denotes the kronecker product and @xmath369 are the transpose of the column vectors @xmath362 and @xmath363 .
let @xmath370 be the random element corresponding to the @xmath371 pattern of @xmath372 $ ] , where @xmath373 $ ] .
using the rank result developed for the rid , it is easy to see that for every @xmath374 $ ] @xmath375;h_j\otimes ( b_1^k)^t)[c_\gamma]\}.\end{aligned}\ ] ] for @xmath183 $ ] , let @xmath339 denote the random increase of rank of @xmath376_{c_{\gamma}}$ ] by adding @xmath377 .
now , consider the stage @xmath336 , where we are going to combine two copies of @xmath318 to construct the matrix @xmath317 .
the the row @xmath321 corresponding to @xmath378 is split into two new rows @xmath337 and @xmath379 which correspond to the row number @xmath338 and the row number @xmath380 of @xmath317 .
@xmath381 similar to the single terminal case , we see that adding @xmath382 increases the rank of the matrix if it increases the rank of the either the first or the second block . in other words , @xmath383 where @xmath384 are the corresponding amount of increase of the rank of the first and second block by adding the @xmath321-the row .
in particular , @xmath282 and @xmath385 are i.i.d .
so are @xmath342 and @xmath343 .
taking the expectation , similar to what did in the single terminal case , we obtain that @xmath386 moreover , one can also show that for @xmath327 $ ] , @xmath387 which together with ( [ plus_part_multi ] ) , implies that @xmath388 .
therefore , @xmath358 is also an erasure process with initial value @xmath197 .
similar to the single terminal case , one can also show that the permutation matrix @xmath351 is not necessary , thus the proof is complete . in this part , we will overview the techniques used to prove the achievability part .
the converse part , given in theorem [ mixture_converse ] , has been proved in appendix [ converse_proof_single ] .
we will give separate constructions for the fully discrete case and the mixture case although the proof techniques used are very similar .
* achievability proof for the mixture case : * we will give an explicit construct of the the measurement ensemble as follows .
let @xmath18 and let @xmath179 .
assume that @xmath180 is a sequence of i.i.d .
nonsingular random variables with rid equal to @xmath48 .
let @xmath181 , where @xmath182 is the hadamard matrix of order @xmath86 . also assume that @xmath184 , @xmath183 $ ] . as we proved in theorem [ single_rid_polarization ]
, @xmath356 is an erasure process with initial value @xmath48 .
we will construct the measurement matrix @xmath90 by selecting all of the rows of @xmath182 with the corresponding @xmath190 value greater than @xmath389 .
therefore , we can construct the measurement ensemble @xmath209 labelled with all @xmath86 that are a power of @xmath288 .
assume that the dimension of @xmath90 is @xmath88 .
it remains to prove that the ensemble @xmath209 is @xmath210-rep with measurement rate @xmath48 .
this will complete the proof of theorem [ achievability_hadamard ] .
we first show that the family @xmath209 has measurement rate @xmath48 .
notice that the process @xmath190 converges almost surely .
thus , it also converges in probability .
specifically , considering the uniform probability assumption , this implies that @xmath390 : i_n(i)\geq \epsilon\ , d(x ) \}}{n}\\ & = \limsup_{n \to \infty } { \mathbb{p}}(i_n \geq \epsilon\ , d(x))\\ & = { \mathbb{p}}(i_\infty \geq \epsilon\ , d(x))=d(x).\end{aligned}\ ] ] it remains to prove that @xmath209 is @xmath210-rep .
let @xmath391 : i_n(i ) \geq \epsilon \
, d(x)\}$ ] denote the selected rows to construct @xmath90 and let @xmath181 be the full measurements .
it is easy to check that @xmath392 . also let @xmath393 $ ] denote all of the indices in @xmath21 before @xmath321 .
we have @xmath394 which shows the @xmath210-rep property for @xmath209 .
* achievability proof for the discrete case : * for the discrete case , the construction of the measurement family is very similar to the mixture case with the only difference that instead of using the erasure process corresponding to the rid , we use the discrete entropy function .
more exactly , in the discrete case , assuming that @xmath181 , we define the following process for @xmath395 $ ] , @xmath396 . in @xcite , using the conditional epi result @xcite , the following was proved .
@xmath397 is a positive martingale converging to @xmath12 almost surely .
similar to the mixture case , we again construct the family @xmath209 by selecting those rows of the shuffled hadamard matrix with @xmath356 value greater than @xmath398 . by a similar procedure ,
it is easy to show that @xmath209 has zero measurement rate .
@xmath399 moreover , assuming that @xmath391 : i_n(i ) \geq \epsilon\ , h(x_1 ) \}$ ] and @xmath393 $ ] , we have @xmath400 which show the @xmath210-rep property for @xmath209 .
the last step is to prove theorem [ universal_mixture ] , namely , to show that for a family of mixture distributions @xmath222 with strictly positive rid , there is a fixed measurement family @xmath209 which is @xmath210-rep for all of the distributions in @xmath222 with a measurement rate vector lying in the rnyi information region of of the family .
the proof is simple considering the fact that the construction of the family @xmath209 in the proof of theorem [ achievability_hadamard ] depends only on the erasure pattern .
also , the erasure pattern is independent of the shape of the distribution and only depends on its rid .
moreover , it can be shown that the erasure patterns for different value of @xmath49 are embedded in one another , namely , for @xmath401 , @xmath402 $ ] . considering the method we use to construct the family @xmath209
, this implies that an @xmath210-rep measurement family designed for a specific rid @xmath49 is @xmath210-rep for any distribution with rid less than @xmath49 .
thus , if we design @xmath209 for @xmath403 , it will be @xmath210-rep for any distribution in the family .
figure [ absorption ] shows the _ absorption phenomenon _ for a binary random variable with @xmath404 .
figure [ polarization ] shows the polarization of the rid for a random variable with rid @xmath405 .
, width=2 ] , width=2 ] in this section , we will give a brief overview of the techniques used to prove the achievability part .
the proof of the converse part is given in appendix [ converse_proof_multi ] .
* acievability proof for the mixture case : * the proof technique is very similar to the single terminal case .
we will define the suitable erasure process and we will use it to construct the desired @xmath210-rep measurement matrices for the multi terminal case . let @xmath234 , @xmath327 $ ] , be a two - terminal memoryless source , where @xmath86 is a power of two .
let @xmath181 and @xmath198
. for @xmath327 $ ] , let us define @xmath184 and @xmath406 . using theorem [ multi_rid_polarization ]
, we can show that @xmath190 and @xmath199 are erasure processes with initial values @xmath48 and @xmath197 polarizing to @xmath193 . the next step is to construct the two terminal measurement ensemble .
let @xmath18 and @xmath179 .
we will construct @xmath407 by selecting those rows of the hadamard matrix , @xmath182 , with @xmath408 .
similarly , @xmath409 is constructed by selecting those rows of @xmath182 with @xmath410 .
it remains to prove that the family @xmath243 labeled with @xmath86 , a power of @xmath288 , and of dimension @xmath411 and @xmath412 is @xmath210-rep with measurement rate @xmath413 . by this construction
, we can achieve one of the corner points of the dominant face of the rate region .
if we switch the role of @xmath24 and @xmath34 we will get the other corner point @xmath414 .
one way to obtain any point on the dominant face is to use time sharing for the two family .
however , it is also possible to use an explicit construction proposed in @xcite , which directly gives any point on the dominant face of the measurement rate region without any need to time sharing . we will just prove the achievability for the corner point @xmath413 .
we first show that the family @xmath243 has measurement rate @xmath413 .
notice that the processes @xmath415 converge almost surely thus , thay converge in probability .
specifically , considering the uniform probability assumption and using a similar technique as we used in the single terminal case , we get the following : @xmath416 : i^x_n(i)\geq \epsilon\ , d(x ) \}}{n}\\ & = \limsup_{n \to \infty } { \mathbb{p}}(i^x_n \geq \epsilon\ , d(x))\\ & = { \mathbb{p}}(i^x_\infty \geq \epsilon\ , d(x))=d(x).\end{aligned}\ ] ] similarly , we can show that @xmath417 . it remains to prove that @xmath243 is @xmath210-rep .
let @xmath418 : i_n(i ) \geq \epsilon \
, d(x)\}$ ] and @xmath419 : j_n(i ) \geq \epsilon \ , d(y|x)\}$ ] denote the selected rows to construct @xmath243 and let @xmath420 and @xmath421 be the full measurements for the @xmath4 and the @xmath5 terminal .
let @xmath422 $ ] and @xmath423 $ ] be the set of all indices in @xmath424 and @xmath425 less than @xmath321 .
we have @xmath426 which shows the @xmath210-rep property for the two terminal measurement family @xmath243 .
* achievability proof for the discrete case : * in the fully discrete case , the construction is very similar to the mixture case with the only difference that instead of using the rid , we will use the entropy .
similar to the single terminal case , we can prove the following .
@xmath192 and @xmath427 are positive martingale converging to @xmath12 almost surely .
we again construct the family @xmath243 by selecting those rows of @xmath182 with @xmath428 and @xmath429 .
similar to the single terminal case , it is easy to show that @xmath243 has measurement rate @xmath430 .
it remains to prove that @xmath243 is @xmath210-rep .
let @xmath418 : i_n(i ) \geq \epsilon \
, h(x)\}$ ] and @xmath419 : j_n(i ) \geq \epsilon \ , h(y|x)\}$ ] denote the selected rows to construct @xmath243 and let @xmath420 and @xmath421 be the full measurements for the @xmath24 and the @xmath34 terminal .
let @xmath422 $ ] and @xmath423 $ ] be the set of all indices in @xmath424 and @xmath425 less than @xmath321 .
we have the following : @xmath431 which shows the @xmath210-rep property for the two terminal measurement family @xmath243 .
the last step is to prove theorem [ universal_mixture ] , namely , to show for a family of mixture distributions @xmath222 , there is a fixed measurement family @xmath243 , which is @xmath210-rep for all of the distributions in @xmath222 with a measurement rate in the ryi information region of the family .
the proof is simple considering the fact that the construction of the family @xmath236 in the proof of theorem [ achievability_hadamard_multi ] depends only on the erasure pattern which is independent of the shape of the distribution and only depends on its rid .
this implies that for any @xmath96 in the rnyi information region of @xmath222 , the designed measurement family @xmath243 is @xmath210-rep @xmath432 .
up to now , we defined the notion of @xmath210-rep for an ensemble of measurement matrices .
this definition is what we call an `` informational '' characterization , in the sense that taking measurements by the ensemble potentially keeps more than @xmath221 ratio of the information of the source .
now , we can ask the natural question that weather this has some `` operational '' implication , in the sense that after having the linear measurements , is it possible to recover the source up to an acceptable distortion ?
in particular , is there a computationally feasible algorithm to do that ? to explain the operational view more , let us give an example from polar codes for binary source compression which has lots of similarities with what we have done . as shown in @xcite , for a binary memoryless source with @xmath433 , for a large block length @xmath20 , there is a matrix @xmath434 , of dimension approximately equal to @xmath435 such that the linear measurement of the source by this matrix over @xmath436 faithfully captures all of the randomness of the source .
this in its own only solves the encoding part of problem without directly addressing the decoding part , namely , it does not imply the existence of a decoder to recover the source from the measurements up to negligible distortion ( error probability ) .
therefore , the operational picture is not complete yet .
fortunately , in the case of polar codes , the successive cancellation decoder ( or other decoders proposed ) fills up the gap and shows that the _ informational _ characterization implies the _ operational _ characterization . for simulations
, we use a unit variance sparse distribution @xmath437 , where @xmath438 is the unit delta measure at point zero , @xmath42 is the distribution of the continuous part and @xmath439 is the rid of the signal .
we use the mse ( mean square error ) as distortion measure .
the simulations are done with the hadamard matrix of order @xmath440 . to build the measurement matrix @xmath66
, we select all of the rows of @xmath182 with highest conditional rid , as stated in [ prooftech : single ] , until we get acceptable recovery distortion .
figure [ ell_1_pt ] shows the phase transition ( pt ) diagram for the @xmath441-minimization algorithm .
the simulations are done with @xmath292 different distributions for @xmath42 : gaussian , laplacian and uniform .
the acceptable recovery distortion is set to @xmath442 .
the recovery is successful for the measurement rates above the plotted curves .
the results show the insensitivity of the pt region to the distribution of the continuous components .
-minimization , width=3 ] -minimization , width=3 ] we also used the amp algorithm to recover the signal , where for simplicity , we only did the simulations for the gaussian case for @xmath42 .
the amp iteration is as follows : @xmath443 where @xmath444 is the linear measurements taken by @xmath66 , @xmath445 is the measurement rate , @xmath446 , @xmath447 and where @xmath448 , with @xmath449 independent of the signal @xmath24 and @xmath450 given by the state evolution equation for amp , is the soft - thresholding function designed for the known distribution of @xmath24 . for initialization , we use @xmath451 and @xmath452 . figure [ amp_pt ] compares the pt diagram for amp and @xmath441-minimization .
although amp , with the thresholding function @xmath453 designed for the known distribution of the signal , performs better than @xmath441-minimization , there is still a gap with the optimal line .
s.haghighatshoar acknowledges mr .
adel javanmard for his helpful comments about the amp algorithm .
e.abbe would like to thank sergio verd for stimulating discussions on the rnyi information dimension .
1 e. cands , j. romberg , and t. tao , robust uncertainty principles : exact signal reconstruction from highly incomplete frequency information , " _ ieee transactions on information theory _ , vol .
52 , no . 2 , pp .
489509 , feb .
donoho , compressed sensing , " _ ieee transactions on information theory _ , vol . 52 , no . 4 , pp . 12891306 , apr .
s. kudekar and h.d .
pfister , `` the effect of spatial coupling on compressive sensing , '' _ in proc .
48th annual allerton conference , _ 2010 , pp .
. f. zhang and h.d .
pfister , `` verification decoding of high - rate ldpc codes with applications in compressed sensing , '' _ ieee transactions on inform .
theory _ , vol .
58 , pp . 50425058 , aug .
dimakis , r. smarandache , p. vontobel , `` ldpc codes for compressed sensing , '' _ ieee transactions on information theory _ , to appear .
m. akcakaya and v. tarokh , `` shannon - theoretic limits on noisy compressive sampling '' , _ ieee transactions on information theory _ , vol .
1 , pp . 492504 , jan .
g. reeves and m. gastpar , sampling bounds for sparse support recovery in the presence of noise , " in _ proceedings of the 2008 ieee international symposium on information theory _ , toronto , canada , jul . 2008 .
s. sarvotham , d. baron , and r.g .
baraniuk , measurements and bits : compressed sensing meets information theory , " _ proceedings of the 44th allerton conference on communication , control , and computing , _ monticello , il , sep .
m. wainwright , information - theoretic bounds on sparsity recovery in the high - dimensional and noisy setting , " in _ proc .
ieee int .
informationtheory _ , nice , france , jun .
w. wang , m.j .
wainwright , k. ramchandran , information - theoretic limits on sparse signal recovery : dense versus sparse measurement matrices , " _ ieee transactions on information theory _ , vol .
56 , no . 6 , pp .
29672979 , jun . 2010 .
d. guo , d. baron , and s. shamai ( shitz ) , a single - letter characterization of optimal noisy compressed sensing , " _ in proceedings of the forty - seventh annual allerton conference on communication , _ control , and computing , monticello , il , oct .
y. wu and s. verd , rnyi information dimension : fundamental limits of almost lossless analog compression , " _ ieee transactions on information theory _ , vol .
56 , no .
37213747 , aug .
y. wu and s. verd , optimal phase transitions in compressed sensing , " _ ieee transactions on information theory _ , vol .
58 , no .
10 , pp .
6241 - 6263 , oct .
donoho , a. javanmard , and a. montanari , information - theoretically optimal compressed sensing via spatial coupling and approximate message passing , " submitted to _ ieee transactions information theory _ , dec .
s. haghighatshoar , e. abbe , e. telatar , adaptive sensing using deterministic partial hadamard matrices , " _ ieee international symposium on information theory _ , pp . 18421846 , jul . 2012 .
s. haghighatshoar , e. abbe , e. telatar , new entropy power inequality for integer - valued random variables , " jan .
2013 , available : http://arxiv.org/abs/1301.4185 f. krzakala , m. mzard , f. sausset , y. sun , and l. zdeborov , statistical physicsbased reconstruction in compressed sensing , " preprint , nov .
available : http://arxiv.org/abs/1109.4424 e. abbe , `` universal source polarization and sparse recovery , '' ieee information theory workshop ( itw ) , dublin , aug .
e. arikan , channel polarization : a method for constructing capacity - achieving codes for symmetric binary - input memoryless channels , " _ ieee transactions inform .
it-55 , pp . 30513073 , jul .
e. arikan , source polarization , " in _ proc .
inform . theory _ , austin , 2010 .
e. arikan and e. telatar , on the rate of channel polarization , " _ ieee international symposium on information theory _ , pp .
14931495 , jul .
e. arikan , `` polar coding for the slepian - wolf problem based on monotone chain rules , '' _ ieee transactions inform . theory _ ,
566 570 , jul .
d. baron , m.f .
duarte , s. sarvotham , m.b .
wakin , and r.g .
baraniuk , an information - theoretic approach to distributed compressed sensing , " _ proceedings of the 43rd allerton conference on communication , control , and computing , _ monticello , il , sept .
d. slepian and j.k .
noiseless coding of correlated information sources , _ ieee transactions inform .
theory _ , vol .
19 , pp .
471480 , jul .
e. j. cands , t. tao , decoding by linear programming " , ieee transaction on information theory , vol.51 , pp .
42034215 , dec . 2005 .
e. j. cands , t. tao , `` near - optimal signal recovery from random projections : universal encoding strategies '' , ieee transaction on information theory , vol .
54065425 , dec . 2006 .
a. rnyi,on the dimension and entropy of probability distributions , " _ acta mathematica hungarica _ ,
1 - 2 , mar .
in this section , we will prove theorem [ mixture_converse ] , which constitutes the converse part and puts a lower bound on the minimum number of linear measurements in order to to keep @xmath210-rep property .
we will prove the following lemmas which will be used repeatedly for other parts .
[ cauchy - binnet ] assume that @xmath60 is a full - rank matrix of dimension @xmath61 , for @xmath454 , and @xmath455 .
then , there exists @xmath456 , |s|=m$ ] such that @xmath457 . as @xmath454 from cauchy - binnet formula we have @xmath458 , |s|=m } \det(\phi_s \phi_s^t)\\ & = \sum _ { s\subset [ n ] , |s|=m } \det(\phi_s)^2.\end{aligned}\ ] ] as all of the terms are positive , there must be a @xmath62 $ ] of size @xmath158 such that @xmath459 which implies that @xmath460 .
[ hqxd1 ] let @xmath24 be a continuous random variable with finite differential entropy and let @xmath461_q$ ] .
suppose @xmath462 is a random element for which the differential entropy and the rid of @xmath24 given @xmath462 are well - defined .
then , we have @xmath463 we have @xmath464 we know that @xmath465 , which implies that @xmath466 has a bounded support at most @xmath467 $ ] .
as the uniform distribution maximizes the differential entropy for a fixed support , we have @xmath468)=0 $ ] .
we also have @xmath469 given @xmath462 , @xmath24 has a well - defined differential entropy , which implies that almost surely for all @xmath470 , @xmath24 conditioned on @xmath462 is a continuous random variable .
therefore , @xmath471 taking the limit as @xmath79 tends to the infinity we get the result . putting @xmath462 equal to null in the lemma [ hqxd1 ] , we get the following corollary .
[ hqxd2 ] let @xmath24 be a continuous random variable with a well - defined differential entropy and let @xmath461_q$ ]
. then @xmath472 and @xmath473 .
[ hqpxd_lb ] let @xmath56 be a sequence of i.i.d . continuous random variables and let @xmath474_q$ ] .
assume that @xmath60 is a full - rank matrix of dimension @xmath61 where @xmath454 and @xmath475 .
suppose @xmath462 is a random element such that the differential entropy of @xmath476 given @xmath462 is well - defined
. then @xmath477 . by lemma [ cauchy - binnet ] ,
there is a @xmath62 $ ] of size @xmath158 such that @xmath478 .
hence , we have @xmath479 where @xmath480 $ ] and @xmath481 .
the final result follows by applying lemma [ hqxd1 ] . without any loss of generality
, we can assume that @xmath209 is a full - rank family , otherwise , we can drop some of the rows of @xmath90 and obtain an equivalent family with lower measurement rate . also , we can assume that the rows of @xmath90 are orthonormal .
otherwise , by gram - schmidt procedure , we can obtain an equivalent family with orthonormal rows .
in other words , there is a lower triangular and invertible @xmath482 matrix @xmath483 such that @xmath484 has orthonormal rows .
as @xmath483 is invertible , it results that @xmath485 thus the equivalent family @xmath486 is also @xmath210-rep and has orthonormal rows , namely , @xmath487 , where we again dropped the dependence of @xmath158 on @xmath86 .
we also represent each @xmath488 $ ] as @xmath489 . from @xmath210-rep assumption , for any @xmath490
there is a @xmath491 such that for @xmath492 @xmath493 where @xmath207_q$ ] .
as we are going to take the limit as @xmath79 tends to infinity , we can drop the negligible terms .
in other words , we have @xmath494 where we used the fact that @xmath495 . for a specific realization @xmath496 ,
let @xmath497 : \theta_i=1\}$ ] and @xmath498\backslash { { c_\theta}}$ ] as introduced before .
then , we obtain @xmath499 where @xmath500_q$ ] and @xmath501_q$ ] denote the component - wise quantization of @xmath502 and @xmath257 .
we also used the fact that @xmath503 let @xmath504_q$ ] and @xmath505_q$ ] .
we consider two cases : first , if @xmath506 , using ( [ mixture4 ] ) , we have @xmath507 second , if @xmath508 , generally , @xmath509 is neither full - rank nor orthonormal . however , we can drop the redundant rows and by using the gram - scmidt procedure , we can create an equivalent orthonormal matrix @xmath510 of dimension @xmath511 with @xmath512 . therefore ,
for this case we obtain @xmath513 where @xmath514 is the variance of @xmath515 . we also used lemma [ hqpxd_lb ] , @xmath516 , and the fact that the gaussian distribution maximizes the differential entropy for a given covariance matrix . combining ( [ mixture5 ] ) and ( [ mixture6 ] ) , and
@xmath517 we obtain @xmath518 which implies that @xmath519 moreover , from ( [ mixture1 ] ) , ( [ mixture2 ] ) and ( [ mixture7 ] ) , we get @xmath520 taking the limit as @xmath79 tends to infinity , we obtain @xmath521 which implies that @xmath522 as @xmath490 is arbitrary , we get the result .
this section is devoted to the proof of theorem [ converse_theorem_multi ] .
this theorem puts constraints on the number of linear measurements we should take from different terminals in order to keep @xmath210-rep property . from @xmath210-rep property ,
we have @xmath523_q,[y_1^n]_q ; & \phi^x_n x_1^n , \phi^y_n y_1^n)\nonumber\\ & \geq ( 1-\epsilon ) h([x_1^n]_q,[y_1^n]_q).\label{first_formula}\end{aligned}\ ] ] similar to the @xmath524 notation that we used in ( [ gamma_notation ] ) for the representation for @xmath180 and @xmath525 , we have @xmath523_q&,[y_1^n]_q ; \phi^x_n x_1^n , \phi^y_n y_1^n ) \\ & \doteq i([x_1^n]_q,[y_1^n]_q ; \phi^x_n x_1^n , \phi^y_n y_1^n,\gamma_1^n)\\ & \doteq i([x_1^n]_q,[y_1^n]_q ; \phi^x_n x_1^n , \phi^y_n y_1^n|\gamma_1^n)\end{aligned}\ ] ] as @xmath526 takes finitely many values , we can obtain the result for a specific realization @xmath527 and then take expectation over all possible realizations . for a specif realization @xmath527 , if some of the components of @xmath528 and @xmath529 are discrete or they are linearly dependent we can drop them . with some abuse of notation ,
let @xmath530 denote the remaining components which have will have dimension @xmath531 and @xmath532 , where @xmath533 and @xmath534 depend on the specific realization @xmath527 .
@xmath535_q,[y_1^n]_q ; \phi^x_n x_1^n , \phi^y_n y_1^n)\nonumber\\ & = h_\gamma(\phi^x_n x_1^n , \phi^y_n y_1^n ) \label{form1}\\ & - h_\gamma(\phi^x_n x_1^n , \phi^y_n y_1^n|[x_1^n]_q,[y_1^n]_q)\nonumber\\ & \preceq - h_\gamma(\phi^x_n x_1^n , \phi^y_n y_1^n|[x_1^n]_q,[y_1^n]_q)\nonumber\\ & \doteq - h_\gamma(q\phi^x_n x_1^n , q\phi^y_n y_1^n|[x_1^n]_q,[y_1^n]_q)\label{form2}\\ & + ( r^x_n + r^y_n ) \log_2(q)\nonumber\\ & \preceq ( m^x_n+m^y_n ) \log_2(q)\nonumber,\end{aligned}\ ] ] where in ( [ form1 ] ) , we used the fact that @xmath536 is upper bounded by the differential entropy of a gaussian random vector with appropriate covariance matrix which vanishes in the limit as @xmath79 tends to infinity . also , in ( [ form2 ] ) , we used lemma [ hqpxd_lb ] to show that @xmath537_q,[y_1^n]_q)\doteq0 $ ]
. therefore , taking the expectation over @xmath526 we obtain that @xmath523_q,[y_1^n]_q ; \phi^x_n x_1^n , \phi^y_n y_1^n)\preceq ( m^x_n+m^y_n)\log_2(q).\end{aligned}\ ] ] we also have @xmath215_q,[y_1^n]_q ) \doteq
n d(x , y ) \log_2(q)$ ] .
therefore , using ( [ first_formula ] ) and taking the limit as @xmath79 tends to infinity , we obtain @xmath538 to prove the other two inequalities , notice that @xmath539_q&,[y_1^n]_q ; \phi^x_n x_1^n , \phi^y_n y_1^n)\nonumber\\ & = i_\gamma([y_1^n]_q ; \phi^x_n x_1^n , \phi^y_n y_1^n ) \nonumber\\ & + i_\gamma([x_1^n]_q ; \phi^x_n x_1^n , \phi^y_n
y_1^n | [ y_1^n]_q)\nonumber\\ & \leq h_\gamma([y_1^n]_q ) + i_\gamma([x_1^n]_q ; \phi^x_n x_1^n|[y_1^n]_q)\nonumber \\ & + i_\gamma([x_1^n]_q ; \phi^y_n y_1^n | \phi^x_n x_1^n,[y_1^n]_q).\label{formula_for_xy}\end{aligned}\ ] ] for the last term , @xmath540_q ; \phi^y_n y_1^n | \phi^x_n x_1^n,[y_1^n]_q)$ ] , we can again assume that we have dropped all of discrete and linearly dependent terms from @xmath529 so that it has a well - defined differential entropy .
thus , we obtain @xmath539_q & ; \phi^y_n y_1^n | \phi^x_n x_1^n,[y_1^n]_q)\nonumber\\ & = i_\gamma([x_1^n]_q ; q\phi^y_n y_1^n | \phi^x_n x_1^n,[y_1^n]_q)\nonumber\\ & = h_\gamma(q\phi^y_n y_1^n| \phi^x_n x_1^n,[y_1^n]_q)\label{formula_for_y1}\\ & - h_\gamma(q\phi^y_n y_1^n| \phi^x_n x_1^n,[y_1^n]_q,[x_1^n]_q)\label{formula_for_y2}.\end{aligned}\ ] ] notice that , for the first term in ( [ formula_for_y1 ] ) , @xmath541_q)&\leq h_\gamma(q \phi^y_n(y_1^n - [ y_1^n]_q)).\end{aligned}\ ] ] it is easy to see that the random vector @xmath542 ) \in [ 0,\frac{1}{q}]^{r^y_n}$ ] has a bounded support independent of @xmath79 thus @xmath543_q)$ ] has an upper bound independent of @xmath79
. therefore , @xmath544_q ) \preceq 0.\ ] ] using a similar argument for ( [ formula_for_y2 ] ) , we have @xmath544_q,[x_1^n]_q ) \preceq 0.\ ] ] assume @xmath260 is a lower triangular invertible matrix , obtained through the gram - scmidt procedure , such that @xmath545 is an orthonormal matrix . then applying lemma [ hqpxd_lb ]
, we obtain that @xmath546_q ) \succeq - \log_2(|\det(l)| ) \succeq 0,\\ & h_\gamma(q\phi^y_n y_1^n| \phi^x_n
x_1^n,[y_1^n]_q,[x_1^n]_q ) \succeq - \log_2(|\det(l)| ) \succeq 0.\end{aligned}\ ] ] this implies that @xmath540_q ; \phi^y_n y_1^n | \phi^x_n x_1^n,[y_1^n]_q)\doteq 0 $ ] .
thus , from ( [ formula_for_xy ] ) , we obtain @xmath547_q,[y_1^n]_q ; \phi^x_n x_1^n , \phi^y_n y_1^n)\\ & \preceq h_\gamma([y_1^n]_q ) + i_\gamma([x_1^n]_q ; \phi^x_n x_1^n|[y_1^n]_q ) . \end{aligned}\ ] ] again if @xmath528 has discrete components or if some of the components are linearly dependent or can be predicted from @xmath548_q$ ] we can drop them . with some abuse of notation
let @xmath528 denote the resulting random vector of dimension @xmath533 .
we have @xmath539_q & ; \phi^x_n x_1^n|[y_1^n]_q)\nonumber\\ & = h_\gamma(\phi^x_n x_1^n|[y_1^n]_q)-h_\gamma(\phi^x_n x_1^n|[y_1^n]_q,[x_1^n]_q)\nonumber\\ & \doteq -h_\gamma(\phi^x_n x_1^n|[y_1^n]_q,[x_1^n]_q)\nonumber\\ & \doteq -h_\gamma(q\phi^x_n x_1^n|[y_1^n]_q,[x_1^n]_q ) + r^x_n \log_2(q)\nonumber\\ & \doteq r^x_n \log_2(q ) \preceq m^x_n \log_2(q),\label{last_formula}\end{aligned}\ ] ] where we used the fact that @xmath549_q)\doteq0 $ ] and from lemma [ hqpxd_lb ] , @xmath550_q,[x_1^n]_q ) \doteq 0 $ ] .
therefore , taking the expectation over @xmath526 and using ( [ first_formula ] ) and ( [ last_formula ] ) , we obtain @xmath551_q,[y_1^n]_q|\gamma_1^n)-h([y_1^n]_q|\gamma_1^n),\\ & \doteq ( 1-\epsilon ) h([x_1^n]_q,[y_1^n]_q ) - h([y_1^n]_q),\end{aligned}\ ] ] which implies that @xmath552 .
therefore , taking the limit as @xmath86 tends to infinity , we get @xmath553 the last inequality in the theorem follows by symmetry . | this paper shows that the rnyi information dimension ( rid ) of an i.i.d . sequence of mixture random variables polarizes to the extremal values of 0 and 1 ( fully discrete and continuous distributions ) when transformed by an hadamard matrix .
this provides a natural counter - part over the reals of the entropy polarization phenomenon over finite fields .
it is further shown that the polarization pattern of the rid is equivalent to the bec polarization pattern , which admits a closed form expression .
these results are used to construct universal and deterministic partial hadamard matrices for analog to analog ( a2a ) compression of random i.i.d .
signals .
in addition , a framework for the a2a compression of multiple correlated signals is developed , providing a first counter - part of the slepian - wolf coding problem in the a2a setting .
rnyi information dimension , polarization , information preserving matrices , analog compression , distributed analog compression , compressed sensing . |
the relative proximity of the andromeda galaxy ( m31 ) and the global perspective from our external vantage point make m31 an excellent laboratory for studying the stellar halos of large galaxies . resolved stellar maps of m31 s halo , assembled over the past decade , have revealed highly complex inhomogeneities , the most striking of which is the giant southern stream ( gss ) , extending @xmath1 away from m31 s center in the southeast direction ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* hereafter i07 ) and falling towards m31 s center with relative radial velocities as high as @xmath2 @xcite .
other significant morphological and kinematic features in the m31 halo include stellar shelves ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* hereafter / ) as well as the recently discovered minor axis `` streams '' .
the gss is especially notable because it offers an opportunity to precisely measure m31 s potential ( * ? ? ?
* ; * ? ? ?
* hereafter ) and provides a view into the most significant local group galaxy disruption in the last gyr .
models detailing the formation of the gss agree remarkably well with most aspects of the observations , and suggest the progenitor had a stellar mass of @xmath3 ( * ? ? ?
our kinematic analysis in f07 finds that seemingly unrelated features like the `` northeast shelf '' and less prominent `` western shelf '' are also the result of the same disruption process , a conclusion supported by independent studies of their stellar populations @xcite .
the observed gss s most striking point of contrast with the models is its asymmetry in the transverse direction . as shown both with photometric samples @xcite and spectroscopic surveys
, its stellar distribution is sharply truncated on the ne side and falls off much more slowly on the sw side .
in addition , the current models do not address the observed stellar population gradients within the gss ( * ? ? ? * ; * ? ? ?
* i07 ) . in this letter
, we show that this structure in the gss can be accounted for if the progenitor hosted a cold , rotating stellar disk , unlike the simple spherical progenitors used in previous simulations .
surprisingly , we find that the disruption of a disk galaxy can also give rise to features similar to the recently discovered arc - like minor - axis `` streams '' , leading to the tantalizing possibility that most of the major inhomogeneities observed in the m31 halo are tidal debris from the same galaxy that caused the gss . in section 2 ,
we briefly describe our model for the progenitor and our @xmath4-body study of its tidal disruption . in section 3 ,
we show results from these simulations , focusing on the transverse density profile of the gss , the metallicity gradient , and arc - like structures that overlap the minor axis .
section 4 summarizes our conclusions .
our simulations are based on the methods worked out in our earlier papers : @xcite , , and .
we use the orbit and potential from table 1 of and their spherical plummer model to represent a non - rotating progenitor . for runs with a disk progenitor ,
we use the same initial position and velocity , but substitute a different initial structure of the satellite . briefly
, our disk models assume the satellite is composed of a bulge and rotating disk of stars . for simplicity
we assume that the dark matter associated with the galaxy has been tidally stripped before the encounter modeled here .
we use a hot exponential @xmath5 disk with mass @xmath6 , radial scale length @xmath7 , and vertical scale height @xmath8 .
we add to this a hernquist bulge of mass @xmath9 and scale length @xmath8 .
we initialize both components with the package zeno written by josh barnes .
we evolve the satellite in m31 s potential starting from 12 evenly spaced orientations of the disk . from this
we select two models displaying particularly good agreement with observational features , referred to here as disks a and b. in a forthcoming paper we will conduct a more systematic survey of possible initial conditions and quantify the debris structure in detail ( fardal et al .
, in preparation ) .
the remaining details of the simulations are the same as those in and .
we set the satellite in motion inbound and slightly past apocenter , minimizing initial transients from m31 s tidal forces .
we run the simulations with the multistepping tree code pkdgrav @xcite .
our simulations include the self - gravity of the progenitor satellite , but ignore dynamical friction , perturbations from the other m31 satellites , and the history of the progenitor prior to the orbit that produces the gss , as justified in our earlier papers .
we stop the simulations 840 myr into the run , which is approximately 650 myr past the initial pericenter .
this time was chosen in to give a reasonable match to the debris structure around m31 , including the radii of the `` shells '' on the e and w sides .
figures [ fig.skymaps]b,d show surface density maps based on the disk a and plummer models , respectively .
both models reproduce the main feature of a stream extending to the se .
they also reproduce the observed line - of - sight distances and velocities along the gss . however , the transverse distribution of gss stars is strikingly different between the two models
disk a displays a much sharper ne edge .
the observed star - count maps @xcite are not directly comparable since they contain both non - gss - related m31 components and non - m31 contaminants and are not explicitly calibrated to stellar surface density , but the morphology of the gss in these maps appears much closer to our disk model .
comparison of the minor - axis contamination to the observations of .
the deimos masks ( rectangles ) are grouped into inner minor - axis masks , outer minor - axis masks , and a single mask ( f135 ) offset from the minor axis .
m31 s center is at ( 0,0 ) .
the inset plots for each group show the ratio @xmath10 of the strength of the gss component to the peak of the gss at the same @xmath11 .
@xmath10 is measured as discussed in the text for disks a and b and the plummer model 840 myr into the runs .
the observational estimates and @xmath12 error bars from gilbert et al .
( in preparation ) are plotted as horizontal solid and dotted lines .
the plummer model clearly contributes too much debris on the minor axis .
, width=288 ] figure [ fig.skymaps]d shows that the plummer model results in a large amount of stars spilling over as far as the se minor axis , located to the ne of the gss .
when compared their keck / deimos spectroscopic data near m31 s se minor axis to this model , they noted much less spillover from the gss than predicted by the model .
gilbert et al .
( in preparation ) has quantified this by dividing the number of stars moving with gss - like velocities on the minor axis to those in the gss core at the same projected radius @xmath11 .
for the nine innermost deimos masks on the minor axis combined , this ratio @xmath13 ; for the three outermost masks on the minor axis , @xmath14 ; and for the mask f135 located somewhat nearer the gss , they find a likely detection of gss material with @xmath15 . in figure [ fig.fieldbars ]
, we compare the density of gss stars in all three @xmath4-body models to these results .
we have selected `` gss '' particles by defining the trend of radial velocity @xmath16 with @xmath11 and then taking stars that fall within @xmath17 of this velocity in the given field .
we also restrict the particles to those actually in the gss s `` shell '' .
we then repeat the procedure for a control field located at the peak of the gss at the same @xmath11 , using a smaller interval @xmath18 as the gss core has a sharper velocity distribution .
clearly the two disk models are in better accord with the observations than the plummer model .
the sharper ne edge and smaller minor - axis contamination of the disk models thus imply that the progenitor was rapidly rotating .
we will explore this argument in more detail in fardal et al .
( 2008 , in preparation ) .
the mean color of gss rgb stars is observed to vary in the transverse direction : the gss is significantly broader in blue than in red stars @xcite .
this is probably due to a metallicity gradient
. quantified the metallicity distribution in two gss - dominated regions , one in the center of the gss and one in a less dense `` cocoon '' region to the sw , and showed that the latter has a lower mean metallicity .
disk galaxies , of course , tend to have metallicity gradients .
therefore it is interesting to see how a plausible gradient in our disk progenitor translates to the metallicity pattern on the sky .
we use a simple parametric model to produce a plausible metallicity gradient in our initial disk model .
we first find the specific orbital energy @xmath19 of each particle .
we then assign it a metallicity using @xmath20}= a_z + b_z \
, \log_{10 } \left [ -e_i/(50 { \,\mbox{km}\,\mbox{s}^{-1}})^2 \right]$ ] , setting @xmath21 and @xmath22 to agree with the results of as explained below .
this produces the metallicity gradient seen in figure [ fig.metals]a .
observational results for the stars in the small disk galaxy m33 are also plotted , with the radius for both galaxies normalized by the disk scale length ; the metallicity pattern of our disk model agrees quite well .
the gss progenitor should perhaps be lower in metallicity than m33 by a few tenths of dex due to its lower inferred mass , but the photometric metallicity measurements probably have systematic uncertainties at this level in any case .
figure [ fig.skymaps]e shows the sky view of the resulting model metallicity pattern .
the gradient along the stream is very weak , but the mean metallicity along the denser central part is clearly higher than in the broad wing to the sw , similar to the pattern seen in m31 s gss . using s figure 27 , we estimate the `` core '' and `` cocoon '' regions ( at @xmath23 kpc ) have mean metallicities of @xmath20}= -0.54 $ ] and @xmath24 , respectively ( mean @xmath20}=-0.51 $ ] was obtained at the gss sharp ne edge by @xcite ) .
figure [ fig.skymaps]e shows `` broad wing '' and `` central gss '' boxes chosen at a similar radius , but better matching the slightly different model stream position . once we set @xmath25 and @xmath26 , the metallicities in these boxes are also @xmath27 and @xmath24 .
the bare fact we can match two metallicities with two parameters is not in itself meaningful , but it is significant that the magnitude _ and sign _ of our initial metal gradient are very reasonable ( fig . [
fig.metals]a ) . figure [ fig.metals]b shows that within each box there is a wide range of metallicities ; the distributions in appear somewhat broader , but given measurement errors and the contributions from other halo components this is not surprising . using their megacam photometric survey of m31 s halo ,
found multiple surface density ridges along the minor axis which they called `` streams '' .
streams c and d ( the two closest to m31 ) form a pair of curving ridges at slightly different orientations , which appear to merge as they approach the survey boundary ( see their fig .
stream c appears to be slightly broader than stream d , and slightly more metal - rich , though not as metal - rich as the gss core / cocoon . from i07 s figure 33 we estimate the mean metallicity of streams c and d to be @xmath28 and @xmath29 respectively .
@xcite suggested these `` streams '' might be shell features from a satellite disruption , similar to the event that created the gss but from a different progenitor .
while studying our overall sample of runs based on 12 disk orientations , we noticed one ( disk b ) containing two curious `` arcs '' crossing the minor axis .
these arcs are clearest at the step 680 myr into the run shown in figure [ fig.skymaps]c .
morphologically , the two arcs somewhat resemble streams c and d , with a fatter southern arc nearly merging into a sharper northern arc . like the observed `` streams '' , neither arc crosses the gss to the sw .
compared to the observed arcs , the simulated arcs are significantly further from m31 s center .
as figure [ fig.skymaps]f shows , the simulated arcs are significantly less rich in metals than the gss . using the same metallicity model as for disk a and the regions defined by boxes in this figure
, the mean @xmath20}$ ] is @xmath30 for the southern arc and @xmath31 for the northern arc .
thus there is considerable if inconclusive evidence that these arcs are close analogues of the `` streams '' in . in our model ,
these two arcs originate from the outer regions of the disk , and are sharp mainly because of the relatively cold velocity field of the disk .
both arcs consist of material that takes a path around m31 s center nearly opposite to the bulk of the progenitor , explaining why they lie so far from the gss .
the large size of our disk is thus crucial ; a compact progenitor resembling m32 , for example , would be unable to produce similar arcs .
the southern arc consists of a group of particles sharing nearly the same energy , and come from fairly far out in the progenitor s disk .
the northern arc consists of particles that lie even further out ( explaining its lower metallicity on average ) , which form a tidal tail during the interaction with m31
. we can not yet explore the full parameter space of the collision for the presence and properties of these arc - like features .
however , we did conduct a few additional runs with changes to the disk mass , radius , and orientation of disk b , finding the arcs were sensitive to the exact input parameters .
thus we will require more theoretical investigation as well as more observational constraints to determine whether the arcs explain some of the minor - axis streams , or are merely a fortuitous similarity .
if the arcs are shown to be related to the gss , they will be a very solid argument for the disk nature of the progenitor , and will place strong constraints on the parameters of the collision .
in summary , several strands of observational evidence suggest that the gss originated from a progenitor with a strong sense of rotation , such as a disk galaxy .
the transverse density profile of the gss is more easily produced by a rotating satellite .
the observed decline in mean metallicity from the central core of the gss to its `` cocoon '' to the sw suggests that the progenitor had a strong radial metallicity gradient , of the sort found mainly in disk galaxies .
furthermore , several observed arcs lying across the minor axis in m31 have very suggestive analogues in one of our runs .
if shown to be related to the gss in the manner suggested by our model , these features would be definite confirmation of a disk - like progenitor .
the notion of a disk galaxy progenitor is somewhat at odds with age measurements of the gss , which suggests little star formation during the last 4 gyr @xcite .
however , the fields used to infer this were placed in the central , metal - rich part of the gss ; it is possible that the progenitor had an age gradient as well as a metallicity gradient , with the older stars on the inside .
age measurements in the gss cocoon would therefore be interesting .
it is also possible that the gss progenitor was more similar to an s0 galaxy than a spiral , perhaps due to stripping of its gas in an earlier phase of its encounter with m31 .
many papers have used metallicity to assess the relationship among various m31 disk and halo features .
our suggestion that the gss progenitor had a strong metallicity gradient means that metallicity can no longer be used as a reliable fingerprint of origin .
this obviously complicates the forensic reconstruction of m31 s merger history . despite this ,
the rapidly growing database on m31 halo structure is a fascinating puzzle , offering unique insight into the life of a typical disk galaxy and the death of its unfortunate former companions .
guhathakurta , p. , et al.2005 , arxiv preprint ( astro - ph/0502366 ) guhathakurta , p. , et al.2006 , aj , 131 , 2497 ibata , r. , irwin , m. j. , ferguson , a. m. n. , lewis , g. , & tanvir , n. 2001 , nature , 412 , 49 ibata , r. , chapman , s. , ferguson , a. m. n. , irwin , m. , lewis , g. , & mcconnachie , a. 2004 , mnras , 351 , 117 ibata , r. , martin , n. f. , irwin , m. , chapman , s. , ferguson , a. m. n. , lewis , g. f. , & mcconnachie , a. w. 2007 , , 671 , 1591 ( i07 ) | the halo region of m31 exhibits a startling level of stellar inhomogeneities , the most prominent of which is the `` giant southern stream '' . our previous analysis indicates that this stream , as well as several other observed features , are products of the tidal disruption of a _ single _ satellite galaxy with stellar mass @xmath0 less than 1 gyr ago . here we show that the specific observed morphology of the stream and halo debris favors a cold , rotating , disk - like progenitor over a dynamically hot , non - rotating one .
these observed characteristics include the asymmetric distribution of stars along the stream cross - section and its metal - rich core / metal - poor sheath structure .
we find that a disk - like progenitor can also give rise to arc - like features on the minor axis at certain orbital phases that resemble the recently discovered minor - axis `` streams '' , even reproducing the lower observed metallicity of these streams .
though interpreted by the discoverers as new , independent tidal streams , our analysis suggests that these minor - axis streams may alternatively arise from the progenitor of the giant southern stream .
overall , our study points the way to a more complete reconstruction of the stream progenitor and its merger with m31 , based on the emerging picture that most of the major inhomogeneities observed in the m31 halo share a common origin with the giant stream . |
processes accompanying ion collisions continue toattract considerable attention due to the advances in accelerator technology .
in particular , a new facility for antiproton and ion research ( fair ) is being constructed based on the gsi helmholtz center for heavy ion research at darmstadt , germany @xcite .
experiments with heavy ions open wide prospects for the verification of quantum electrodynamics in strong magnetic fields .
the generation of an electron - positron pair upon collisions of ions is one of the most interesting processes .
this process was considered for the first time by landau and lifshitz @xcite and was later studied in detail in the high energy range as well as for low - energy collisions ( see , for example , @xcite and references therein ) . in the latter case
, a quasi - molecule can be formed . the magnetic field produced by moving ions
can reach and even exceed the critical quantum - electrodynamics value of @xmath0 g even for ion energies on the order of the coulomb barrier .
nevertheless , the effect of the magnetic field is usually disregarded in accordance with the conclusions drawn in @xcite . however , as shown in @xcite , the magnetic field can produce a substantial effect due to the interaction with the generated pair .
the mechanism of such an interaction is analogous to magnetic field line freezing into a plasma ( this effect is well known in the plasma physics and astrophysics ) . as a result ,
the lifetime of the magnetic field increases significantly and can considerably exceed the nuclear time of flight .
an observable indication of this effect is the presence of resonances in the spectrum of generated pairs , which are typical of processes occurring in a magnetic field .
it is interesting to note that anomalous peaks were observed in gsi experiments on collisions of heavy ions @xcite , although the reproduction of the results unfortunately remains problematic .
we believe that the main features of this effect can be investigated using the following model . the formation of a pair can be described as a photoproduction by two photons using the well - known equivalent - photon approximation .
the corrections corresponding to the interaction with the magnetic field can be taken into account in the context of the furry picture ( i.e. , using the electron wavefunctions in a magnetic field ) .
it should be noted that this technique is generally analogous to the bethe
maximon method used for describing collisions of ions with relativistic energies @xcite .
therefore , in the simplest case , the problem can be reduced to two - photon generation of a pair in a magnetic field .
this problem is also of independent importance in astrophysics . in particular , the effectiveness of generation of an electron
positron plasma in the magnetospheres of pulsars by competing one- and two - photon processes was considered in @xcite . the process of generation of an electron
positron pair by two photons in a strong magnetic field was studied in @xcite in the case of a head - on collision of photons along the magnetic field line .
this process was analyzed in @xcite in the nonresonant case , when the energy of each of the photons is insufficient for generating a pair in a one - photon process .
the mean free path of a high - energy photon propagating through a photon gas along magnetic field lines was calculated in @xcite . in this study
, we consider the effect of the resonant production of an electron
positron pair by two polarized photons in a magnetic field on the excited landau levels taking into account the spins of the particles .
we calculate the resonant cross section for arbitrary polarizations of particles in the case when the electron and the positron occupy arbitrary low landau levels .
we use the relativistic system of units ( @xmath1 ) .
the expression for the amplitude of two - photon production of an electron
positron pair in a magnetic field has the form @xmath2,\end{gathered}\ ] ] where @xmath3 are the potentials of plane waves @xcite and @xmath4 are the solutions to the dirac equation in a magnetic field for the electron and positron ( primed values are functions of @xmath5 ) .
we direct the @xmath6 axis along the magnetic field and choose the vector potential in the form @xmath7 .
then the electron wavefunctions describe the states with certain values of energy and momentum @xcite .
the energy eigenvalues are given by @xmath8 where @xmath9 is the electron mass , @xmath10 is the number of the landau level , and @xmath11 is the magnetic field strength in units of the critical field : @xmath12 the electron propagator in the given basis has the form @xcite @xmath13 @xmath14 where @xmath15 , @xmath16 are the dirac gamma matrices , @xmath17 the argument of hermite functions @xmath18 in eq .
( [ gh ] ) has the form @xmath19 it should be noted that propagator ( [ g ] ) was also obtained independently in @xcite figure [ diagrams ] shows the feynman diagrams corresponding to amplitude ( [ sfi ] ) .
the amplitude was obtained in explicit form in @xcite .
the kinematics of the process is determined by the law of conservation of energy and the @xmath6 component of the momentum , @xmath20 ( here and below , superscripts `` @xmath21 '' and `` @xmath22 '' correspond to the positron and electron , respectively ) .
it can easily be verified @xcite that the threshold frequencies and momenta of photons satisfy the condition @xmath23 it can be seen that this conditions can not be satisfied if both photons move parallel to the field in the same direction . in this case , the left - hand side of eq .
( [ th ] ) is zero , while the right - hand side is always greater than @xmath24 .
it should be noted that the lorentz transformations along the magnetic field do not change the field itself .
therefore , without loss of generality , we can eliminate the longitudinal momentum of the photons by appropriately choosing the reference system so that @xmath25 in addition , we consider the process in the ultra - quantum - mechanical , or low landau levels ( lll ) approximation , in which the following conditions hold : @xmath26 we introduce detuning @xmath27 from the threshold , @xmath28 where we have introduced the notation @xmath29 .
we assume that the detuning is on the order of the spacing between the landau levels @xmath30 . taking into account conditions ( [ lll ] ) ,
we can then write the momentum of the generated particles as @xmath31
when quantities @xmath32 and @xmath33 in propagator ( [ g ] ) satisfy the relativistic relation between the energy and momentum of a particle in a magnetic field , resonance takes place .
the resonance condition is the equality to zero of the denominator of the green s function : @xmath34 quantities @xmath32 and @xmath33 are defined in accordance with the laws of conservation at the vertices of the diagram .
equations ( [ res ] ) define two resonance conditions . we will use conditions ( [ lll ] ) and expression ( [ pzlll ] ) for the momentum of finite particles . in this case ,
resonance frequencies of photons in the lowest approximation are given by @xmath35 @xmath36 for the `` @xmath21 '' and `` @xmath22 '' signs in eq .
( [ res ] ) , respectively .
it should be noted that the intermediate particle is the electron in the former case and the positron in the latter case .
it is well known that a second - order process in the fine - structure constant at resonance can be depicted as a sequence of first - order processes . in our case , this is the generation of a pair by a single photon and absorption of the photon in the magnetic field .
it follows from eqs .
( [ resw1 ] ) and ( [ resw2 ] ) that one of the photons ( the hard one ) forms a pair and should have an energy exceeding the one - photon generation threshold .
the second ( soft ) photon is absorbed by the electron , and its frequency should be equal to the energy of transition between landau levels ( i.e. , multiple to cyclotron frequency @xmath37 ) .
the resonance conditions in the feynman exchange diagram can be obtained from expressions ( [ resw1 ] ) and ( [ resw2 ] ) by interchanging photon indices @xmath38 . in the general case
, number @xmath39 also changes @xmath40 .
we can obviously formulate the conditions in which resonance is possible for two diagrams simultaneously .
indeed , equating the frequencies of soft photons in the direct and exchange diagrams , we obtain @xmath41 let us formulate the condition for interference of resonances in a more convenient form .
we can write the energy of the soft photon in resonance as @xmath42 then , simultaneous resonance of the amplitudes is possible for values of @xmath39 and @xmath43 given by @xmath44 obviously , interference of resonances is impossible if @xmath45 or @xmath46 .
indeed , the soft photon in resonance is absorbed by different particles ( electron and positron ) in the direct and exchange diagrams . if the photon energy is higher than the energy of one of the particles , the resonance condition for the corresponding amplitude is violated .
henceforth , we will mark the quantities pertaining to the soft photon with the subscript `` 1 '' .
in addition , we confine our analysis to the case when resonance occurs only in one diagram . for definiteness , we set @xmath47 . in the first nonvanishing approximation in @xmath11 ,
the resonance frequencies are independent of the angles of incidence of photons .
calculation of the next correction in @xmath11 gives @xmath48 where @xmath49 is the polar angle of the first photon , and the sign in the first expression coincides with the sign @xmath50 .
it should be noted that the hard resonant photon must be directed almost perpendicularly to the magnetic field in view of the choice of reference system ( [ k0 ] ) . using relation ( [ resw1 ] ) ,
we obtain @xmath51 figure [ t_w1 ] shows the dependences of resonance frequencies of photons , which are exact in @xmath11 . +
let us calculate the resonant cross section of the process using expressions ( [ resw1 ] ) for the resonance frequencies and assuming that conditions ( [ lll ] ) is satisfied .
it is well known that the differential cross section of the process has the form @xmath52 where @xmath53 are the intervals of final states of the electron and positron , @xmath54 @xmath55 and @xmath56 are the area and volume of the normalized wavefunctions for the electron and photon , @xmath57 , and @xmath58 is the angle between the directions of the photons . substituting the explicit form of the wavefunctions into expression ( [ sfi ] ) and retaining only the first terms in the expression in small parameter @xmath11 , we obtain @xmath59 where @xmath60 is the fine - structure constant , and indices @xmath61 denote the signs of the spin projections of the electron and position , respectively .
the spin - dependent factors have the form @xmath62 @xmath63 , \end{aligned}\ ] ] @xmath64 @xmath65 .\end{aligned}\ ] ] here , the following notation has been introduced : @xmath66 @xmath67 @xmath68 @xmath69 are the polarization vectors of photons and @xmath70 are their angles of azimuth . in resonance , we have @xmath71 , and the denominator of expression ( [ g ] ) vanishes .
to get rid of the divergence , we introduce width @xmath72 of the intermediate state in accordance with the breit wigner rule @xcite : @xmath73 in calculating the total cross section , the integrals with respect to @xmath74 and @xmath75 can easily be evaluated using the properties of the @xmath76 function . however , expression ( [ dsigma ] ) is independent of variable @xmath77 ; therefore , the result of integration with respect to @xmath78 is factor @xmath79 @xcite . in this case , the expression for the cross section acquires the term @xmath80 in the form proposed in @xcite : @xmath81 finally , after simple computations , the resonant cross section of the process assumes the form @xmath82 , \ ] ] @xmath83 , \ ] ] @xmath84 - 2nn \xi_1\xi_1 s^2 \right\ } , \end{aligned}\ ] ] @xmath85 , \ ] ] where @xmath86 , @xmath87 , @xmath88 , and @xmath89 here , @xmath90 denotes the stokes parameters of the hard photon , @xmath91 are these parameters for the soft photon , and the first and second subscripts in expressions ( [ smp])([spm ] ) indicate the sign of the electron and positron spin projections , respectively .
figure [ s_h ] shows the dependences of the resonant cross section of photoproduction on the magnetic field strength for various spin projections of the particles . it can be seen from expressions ( [ smp])([spm ] ) that the cross section has the highest order of magnitude for the production of a pair with spin directions @xmath92 and @xmath93 as in the case of one - photon generation @xcite .
this spin state corresponds to the minimal energy of interaction of magnetic moments of particles with the magnetic field .
the reversal of the spin projection of each particle reduces the cross section by an order of magnitude in @xmath11 .
the strong dependence of the cross section on the polarization of the hard photon is also worth noting . in particular , for the normal linear polarization , we have @xmath94 , and cross section @xmath95 vanishes .
it should be noted , however , that for correct comparison of cross section @xmath95 with other cross sections , it is necessary to calculate it with corrections in @xmath11 of the corresponding order of magnitude . in the case of the linear polarization of photons , expressions ( [ smp])([spm ] )
are symmetric relative to the polar angle .
l. d. landau and e. m. lifshitz , _ course of theoretical physics _ , volume 4 : v. b. berestetskii , e. m. lifshitz , and l. p. pitaevskii , _ quantum electrodynamics _
( nauka , moscow , 1989 ; pergamon , london , 1994 ) .
a. i. akhiezer and v. b. berestetskii , _ quantum electrodynamics _
( wiley , new york , 1965 ; nauka , moscow , 1981 ) .
p. i. fomin and r. i. kholodov , ukr .
44 * , 1526 ( 1999 ) .
a. v. kuznetsov and a. a. okrugin , int . j. mod .
phys . a * 26 * , 2725 ( 2011 ) . | we consider the resonant generation of an electron
positron pair by two polarized photons to arbitrarily low landau levels .
the resonance occurs when the energy of one photon exceeds the one - photon generation threshold , and the energy of the other photon is multiple to the spacing between the levels .
the cross section of the process is determined taking into account the spins of particles .
the order of magnitude of the cross section is the highest when the magnetic moments of the particles are oriented along the magnetic field .
_ j. exp .
theor . phys . *
121*(5 ) 813 ( 2015 ) ; _ * doi : * 10.1134/s1063776115110126 |
the low - frequency radio sky spectrum between 50 and 200 mhz is a key area of interest because experiments seeking to detect redshifted 21 cm radiation from neutral hydrogen gas during the epoch of reionization ( eor ) and earlier eras of first light and x - ray heating must subtract astrophysical foregrounds in these frequencies to high - precision . at these frequencies ,
the sky is mostly dominated by galactic synchrotron radiation but also contains contributions from supernova remnants and extragalactic radio sources .
there are two approaches to detecting the redshifted 21 cm signal .
the first approach is targeted by interferometeric arrays , such as the murchison widefield array ( mwa ; @xcite ) , the precision array to probe the epoch of reionization ( paper ; @xcite ) , the hydrogen epoch of reionization array ( hera ; @xcite ) , the low - frequency array ( lofar ; @xcite ) , and the low - frequency aperture array component of the square kilometer array ( ska ; @xcite ) .
these arrays aim to measure spatial fluctuations in the sky brightness temperature on arcminute and degree scales resulting from variations in the density , ionisation , and temperature of the intergalactic medium ( igm ) above redshift @xmath7 ( below 200 mhz ) .
the second approach , on the other hand , exploits the bulk properties of the high - redshift igm that yield a global ( monopole ) contribution of redshifted 21 cm signal to the all - sky radio background .
this approach is being pursued by the experiment to detect the global eor signature ( edges ; @xcite ) , broadband instrument for global hydrogen reionisation signal ( bighorns @xcite ) , sonda cosmolgica de las islas para la deteccin de hidrgeno neutro ( sci - hi @xcite ) , large - aperture experiment to detect the dark age ( leda @xcite ) , shaped antenna measurement of background radio spectrum ( saras @xcite ) , and the dark ages radio explorer ( dare ; @xcite ) .
both methods require careful and accurate subtraction of the sky foreground since the 21 cm signal is three to five orders of magnitude below the foreground .
although observations of the low - frequency radio sky were among the earliest in radio astronomy @xcite , detailed knowledge of the spectral properties of diffuse emission remains lacking .
the @xcite all - sky map at 408 mhz , which was compiled from data taken in the 1960s and 1970s , is still the cornerstone for foreground templates extrapolated across many frequencies in cosmic microwave background ( cmb ) and redshifted 21 cm analyses .
improvements to the map have occurred throughout the years , including efforts by @xcite , and @xcite that focused on removing point sources and destriping the original data .
@xcite combined the haslam 408 mhz map and several other surveys between 10 mhz and 94 ghz to build a global sky model ( gsm ) , which provides spectral properties of the diffuse emission of the entire sky . in this model at 150 mhz , the spectral index outside the galactic plane is @xmath8 and in the galactic plane ( in a narrow band ) increases to @xmath9 with peaks as high as @xmath10 .
both the haslam map and the gsm have become widely used and cited in both simulations and data reduction pipelines for redshifted 21 cm observations @xcite .
@xcite created an all - sky temperature map at 45 mhz based upon the surveys of @xcite and @xcite .
in addition , they produced an all - sky galactic spectral index map based upon two frequency points by using their 45 mhz and the haslam 408 mhz map after corrections for zero - level , extragalactic non - thermal emission , and cmb factors . at higher frequencies , observations of galactic radio emission between 3 and 90 ghz
have been made by the absolute radiometer for cosmology , astrophysics , and diffuse emission i and ii ( arcade i and ii ) sky surveys @xcite .
this survey constrained models of extragalactic emission and suggested that models of electric dipole emission from spinning dust particles were a possible explanation for excess emission seen at 1 cm wavelengths .
they find that , at most , three parameters ( reference brightness temperature , spectral index , and curvature ) are necessary to model their wide - band data combined with external radio measurements between 22 mhz and 1.4 ghz , and from wmap at 23 ghz @xcite .
recently , the new redshifted 21 cm arrays have begun yielding the first large , multi - spectral maps sensitive to diffuse structures in the low - frequency sky , including recent surveys by mwa gleam @xcite and lofar @xcite .
gleam will scan the entire sky south of @xmath11 between 72 and 231 mhz , and the lofar survey will scan 100 sq degrees in the northern sky centred on @xmath12 between 30 and 160 mhz .
the lofar survey s primary purpose is to enable automated processing by providing an _ a priori _ sky model , which also serves as an aid to foreground removal for eor detection .
similarly , gleam s survey will serve a myriad of uses , one of which again will be foreground removal in eor searches . the edges instrument is able to provide a unique measurement of the absolute sky brightness temperature averaged over large spatial scales on the sky due to its wide beam .
@xcite found that the spectral index , @xmath13 , of diffuse emission , defined as @xmath14 , was @xmath15 at high - galactic latitudes in the frequency range 100 - 200 mhz .
those measurements were taken using three days of data ; two of which used a n - s orientation of the antenna s excitation axis and one with an e - w orientation . by combining their results with the haslam sky map at 408 mhz , they were able to show that @xmath16 at high galactic latitudes .
in this paper we present new observations taken over a span of 240 nights from july 2015 through march 2016 with the latest version of the edges instrument that deploys an improved antenna with better chromatic performance and a new high - precision receiver calibration approach .
these advancements enable us to improve on our earlier measurements of the spectral index of the diffuse low - frequency radio emission and extend our coverage to all sidereal times at a constant declination of @xmath0 .
the paper is organized in the following manner . in section 2
we describe the instrument and calibration details . in section 3
we present details of the data collected and chromatic beam corrections .
section 4 presents and discusses the spectral index results and comparisons to values predicted by relevant sky models .
the edges experiment is deployed at the murchison radio - astronomy observatory ( mro ) in western australia ( -26.7@xmath17 , + 116.6@xmath17 ) .
the experiment consists of two scaled - replica instruments spanning neighboring frequency bands of 50 - 100 mhz and 90 - 190 mhz .
each instrument consists of a single dipole - based antenna and receiver .
extensive modeling and simulation is performed on all instrument components , along with calibration measurements conducted both in the laboratory prior to instrument deployment and on site following deployment , in order to meet the low systematic instrumental error requirements for redshifted 21 cm observations @xcite . for the results reported here
, we used the high - band instrument , hence we focus our discussion on that instrument . in this section ,
we review the instrument and the primary calibration steps .
the high - band instrument employs a single broadband `` blade '' dipole antenna sensitive to wavelengths in the range @xmath18 m ( @xmath19 mhz ) .
the antenna is ground - based and zenith pointing .
the antenna consists of two rectangular planar panels that are each 63 @xmath20 48 @xmath21 and are placed 52 cm above a ground plane .
the metallic ground plane is formed using a 5.35 @xmath20 5.35 m@xmath22 aluminum plate underneath the antenna , with four wire mesh extensions ( each 2 @xmath20 5 m@xmath22 ) that combine to form a `` plus '' shape .
the antenna uses a roberts balun @xcite with a small shield at the bottom ( fig .
[ fig : antenna_images ] ) .
the blade antenna replaces the older `` fourpoint '' antenna design used in prior edges studies and yields better chromatic performance by introducing less spectral structure into the measurement @xcite .
the full width at half maximum beamwidth of the antenna at 150 mhz is 72@xmath17 parallel to the axis of excitation and 110@xmath17 perpendicular to the axis .
the features of the antenna are summarized in table [ tab : antennas ] .
the antenna is connected to a receiver placed underneath the ground plane in a temperature - regulated enclosure kept at @xmath23c .
the receiver amplifies and conditions the signal before passing it through a 100 m cable to a backend unit that further amplifies it and sends it to a pc - based 14-bit 400-ms / s analog to digital converter ( fig .
[ fig : block_diagram ] ) . blocks of 65536 voltage samples are then fourier transformed to obtain spectra of 32768 points below 200 mhz with 6.1 khz resolution .
the receiver utilizes laboratory calibration prior to deployment , augmented with a three - position hot / cold / antenna calibration switching scheme for stability in the frontend during operation , to achieve an absolute accuracy of @xmath24% in measured antenna temperature @xcite .
the calibration quantities are measured as a function of both temperature and frequency .
the impedance match of the antenna can be measured _ in situ _ by switching a vector network analyzer ( vna ) into the electrical path . to propagate the calibration to the sky temperature measurement
, we must apply several corrections for losses that occur in the antenna before the receiver .
the losses include ground plane loss , balun loss , and antenna panel loss .
we also make an estimate and correct for the actual temperature of the low - noise amplifier ( lna ) which may deviate from the nominal 25@xmath17c due to temperature gradients .
ground plane loss originates from our finite ground plane , allowing a small amount of signal ( @xmath25 ) from below the horizon to be received from the antenna .
these losses are modeled through cst ( computer simulation technology ) microwave studio and through feko ( feldberechnung fr kper mit beliebiger oberflche ) e&m simulations .
the balun loss is due to minor resistive impedance in the brass and copper - plated balun hardware and also to the resistivity and dielectric properties of the connector to the receiver .
this effect is estimated from analytical models .
finally , losses in the antenna panels are due to finite conductivity and estimated with feko .
adjustments are made to the measured spectrum to counteract these effects in processing .
l l parameter & value + 3 db beamwidth @xmath26= 0@xmath27 at 150 mhz & 72@xmath27 + 3 db beamwidth @xmath26=90@xmath27 at 150 mhz & 110@xmath27 + height above ground plane & 52 cm + panel width & 63 cm + panel length & 48 cm + solid ground plane&5.35 @xmath20 5.35 m@xmath22 + mesh ground plane extensions&2 @xmath20 5 m@xmath22 + & + & + & + & +
we measured the sky temperature using edges over a period of 240 days starting with day 207 ( july 26 ) of 2015 .
we determined the spectral index by fitting the calibrated data to the two parameter equation @xmath28 where the two fitting parameters , @xmath1 and @xmath29 , are the spectral index and the brightness temperature at 150 mhz less @xmath30 , respectively .
the cmb temperature , @xmath30 , is 2.725 k , and @xmath31 is 150 mhz .
we find that this equation yields lower rms fitting errors than the log form of the equation : @xmath32 data collection was occasionally interrupted due to a variety of factors such as power outages and inclement weather ( table [ tab : missing_data ] ) .
we exclude daytime data since effects from the sun and the ionosphere @xcite can interfere with the observation of the astronomical sky foreground .
sporadic data points affected by radio - frequency interference ( rfi ) are flagged and excised .
the data are averaged and binned into 72 timeslots of 20 minutes in lst and into 250 frequency bins of 400 khz from 90 to 190 mhz . for an antenna on the surface of the earth pointed toward zenith , the pointing declination of its primary axis corresponds to the latitude of the antenna s deployment site , and the pointing right ascension corresponds to the lst at the site . as the earth rotates , the antenna pointing changes direction , altering the mapping of its beam pattern onto sky coordinates . because the antenna s beamwidth is wide by design , we sample the average sky temperature weighted by the beam ( see figs . [
fig : beam_width ] - [ fig : galactic_beams ] ) . at low frequencies , the earth
s ionosphere can distort the appearance of the astronomical sky and , in principle , provide an additional chromatic effect .
@xcite used the lofar low frequency antenna beam to evaluate the chromatic effects of the ionosphere , in particular due to refraction . to minimize these effects , we restrict our data analysis to times when ionospheric perturbations are minimal by excluding data when the sun s elevation is -10@xmath27 or higher .
l l c day numbers&year&span + 208 to 209 & 2015&2 + 212 to 214 & & 3 + 244 & & 1 + 246 to 249 & & 4 + 263 to 264 & & 2 + 290 & & 1 + 342 & & 1 + 303 to 309 & & 7 + 20 to 26&2016&7 + 54&2016&1 + total days & & 29 + in the context of measurements of the sky spectrum , an ideal antenna beam would be independent of frequency .
however , as shown in @xcite , realistic antennas suffer chromatic ( frequency - dependent ) effects that couple angular structure in the sky to spectral structure in the measurement .
thus , before we can analyze the spectral properties of the radio sky to high - precision , we must correct for the chromatic effects of the edges antenna .
[ fig : beam_width ] shows a snapshot of the beam s directivity at 150 mhz projected upon the haslam sky map when located at -26.7 latitude and pointed towards the zenith at lst = 13 h. when the directivity of the beam changes at other frequencies , affected locations in the sky will contribute stronger or weaker to the antenna temperature .
these chromatic beam effects are significant especially with the galactic centre overhead where the rms fitting errors to a two parameter equation can exceed 9 k for an uncorrected spectrum .
@xcite chose to report the spectral index using a region in the sky which would minimize chromatic beam effects and chose lst = 2.5 h. fig .
[ fig : galactic_beams ] shows that this is true for the present blade beam , but even in this region , beam effects must be taken into account .
compared to other antenna choices , we have significantly reduced this effect by using the blade design , which has the benefits of minimal chromaticity and is conducive to precision beam modeling through simulations @xcite . to further reduce the chromatic beam effect in measured spectra
, we use simulated beam maps , along with a sky model , to generate a chromaticity correction factor for the beam , which can reduce the rms fitting errors to under 5 k. we model the antenna from 90 to 190 mhz in steps of 1 mhz with numerical time - domain electromagnetic simulations using cst microwave studio .
we then use these beam models , along with the haslam 408 mhz map , to compute a correction factor as a function of frequency for each lst in our observations . to scale the haslam map from 408 mhz to our frequency range
, we first subtract the @xmath33 temperature from the map , scale its brightness temperature to 150 mhz using a constant spectral index of @xmath34 and then add @xmath33 back in . we convolve the scaled haslam map with the beam directivity at each frequency modeled .
the beam factor ratio is formed by dividing the modeled antenna temperature at each frequency by the antenna temperature at 150 mhz .
we implicitly assume that the spectral index does not vary significantly from its value at 150 mhz in our frequency range of 90 to 190 mhz .
the beam correction factor is given by : @xmath35 where @xmath36 \left ( \frac{150}{408 } \right ) ^{-2.5 } + t_\text{cmb } , \ ] ] @xmath37 is the beam directivity , @xmath38 is the haslam sky map , @xmath30 is 2.725 k , and @xmath39 is the frequency .
this ratio is the chromatic beam factor that we apply to the measured spectrum as a divisor to counteract the under or over response to the sky .
we repeat this for all 72 time steps in lst to form a two - dimensional beam factor array ( see fig .
[ fig : beam_factor ] ) .
latitude with the zenith at 13 h lst .
the beam is purposefully wide to capture the sky averaged global eor signal . ]
latitude shown for lst values of 18 h and 2.5 h ( dark trace and light blue trace respectively ) .
the beam at 18 h lst captures much of the central region the galactic plane , which is high in spatial structure , while at 2.5 h it captures mostly structureless sky , minimizing chromatic beam effects . ]
from day number 207 in 2015 to day 82 in 2016 , data were fit separately per day for each lst interval .
the parameters @xmath29 , @xmath1 , and the rms error to the fit were extracted using equation ( [ eq : power_law_1 ] ) .
the spectra were fit for the two cases of using or not using beam chromaticity correction .
the values of @xmath29 , @xmath1 , and the rms error show excellent stability over time as can be seen in the waterfall plots of figs .
[ fig : waterfalls_two_params]-[fig : waterfalls_two_params_rms ] and confirmed by the low standard deviation in data scatter as seen in the @xmath1 and @xmath29 averages of fig .
[ fig : grouped_averages_two_params ] , where @xmath4 < 0.003 and @xmath40 < 5 k. @xmath29 is also insensitive to beam chromaticity as the average beam corrected and uncorrected values differ by no more than 0.2% due to the low chromaticity of the blade beam .
the daytime data ( sun above @xmath41 elevation ) are not shown , yielding the blank band in the plot .
the importance of including the beam correction factor is most apparent in the average @xmath1 plot of fig .
[ fig : grouped_averages_two_params ] . with
the correction applied , we find @xmath42 for lst values between 0 and 12 h. between 12 and 24 h it features a flattening that peaks at @xmath43 at 17.7 h. it is clear in fig . [
fig : grouped_averages_two_params ] that beam correction has a significant impact on the recovered spectral index . without correction , the beam effects mask the flattening that is expected near the galactic centre @xcite .
we also investigated fitting the spectra to a three parameter model , adding @xmath44 as a third term @xmath45 ^ 2 .\ ] ] although the rms fitting errors were smaller and @xmath46 was typically relatively small , < 0.1 , we conservatively assume that the uncertainty on gamma due to systematic effects dominates fitted values .
hence , we save further three parameter analysis to future work . and ( bottom ) the sky temperature at @xmath31=150 mhz , @xmath29 .
rfi and other spurious signals were purged from the spectra before performing a fit to equation [ eq : power_law_1 ] .
the data were binned into 400 khz wide bins from 90 to 190 mhz and the beam adjustment factor was applied .
both the @xmath1 and the @xmath29 graphs show excellent stability from day to day as the data collection ran for 240 days from day 207 2015 to day 82 2016.,title="fig : " ] and ( bottom ) the sky temperature at @xmath31=150 mhz , @xmath29 .
rfi and other spurious signals were purged from the spectra before performing a fit to equation [ eq : power_law_1 ] .
the data were binned into 400 khz wide bins from 90 to 190 mhz and the beam adjustment factor was applied .
both the @xmath1 and the @xmath29 graphs show excellent stability from day to day as the data collection ran for 240 days from day 207 2015 to day 82 2016.,title="fig : " ] ) .
rfi and other spurious signals were purged from the spectra before performing the fit .
the data were binned into 400 khz wide bins from 90 to 190 mhz and the beam adjustment factor was applied .
the rms error also shows excellent stability from day to day as the data collection ran for 240 days from day 207 2015 to day 82 2016 . ] , @xmath29 , and rms error for each lst value taken over days for which nighttime data was available .
averages are shown for the processing done with and without beam correction .
beam correction reveals the dip in @xmath1 near the galactic centre , where the variations of @xmath1 without beam correction are dominated by the chromatic beam effects .
the small data scatter in @xmath1 further confirms the uniformity from day to day seen in the waterfall graphs .
@xmath29 min and max values , 257 k and 842 k , are annoted on the chart.,title="fig : " ] , @xmath29 , and rms error for each lst value taken over days for which nighttime data was available .
averages are shown for the processing done with and without beam correction .
beam correction reveals the dip in @xmath1 near the galactic centre , where the variations of @xmath1 without beam correction are dominated by the chromatic beam effects .
the small data scatter in @xmath1 further confirms the uniformity from day to day seen in the waterfall graphs .
@xmath29 min and max values , 257 k and 842 k , are annoted on the chart.,title="fig : " ] , @xmath29 , and rms error for each lst value taken over days for which nighttime data was available .
averages are shown for the processing done with and without beam correction .
beam correction reveals the dip in @xmath1 near the galactic centre , where the variations of @xmath1 without beam correction are dominated by the chromatic beam effects .
the small data scatter in @xmath1 further confirms the uniformity from day to day seen in the waterfall graphs .
@xmath29 min and max values , 257 k and 842 k , are annoted on the chart.,title="fig : " ] as compared to the measurement derived curve , but the guzmn curve matches the measured data more closely . ] we evaluate and correct for sources of systematic errors to the calibration process : lna temperature gradients , ground loss , antenna loss , and beam correction uncertainty .
we maintain the receiver at @xmath23c with forced airflow and active cooling / heating , but the temperature of the lna s electronic components may differ slightly from the reading at the main temperature sensor .
a second sensor is used to capture potential gradients , which remain below @xmath47c during operation .
this temperature difference has a small effect on the calibration values that are used to adjust the antenna spectrum .
simulations showed the impact on @xmath1 to be no more than 0.005 if we did not include the adjustment . in the field
, we use a @xmath48 m@xmath22 solid metallic ground plane made from a central square of solid aluminum augmented by @xmath49 m@xmath22 sheets of wire mesh .
the imperfect ground of the earth serves as the ground plane where there is no metal .
given the limited size of the metal ground plane , a small fraction of the antenna beam will pickup the temperature of the ground .
the effect of the finite ground plane on ground pickup / loss was simulated in feko and cst .
in feko we used a greens function to implement a finite ground plane modeled over the soil .
we used values for the dielectric constant of 3.5 and conductivity of 0.02 s / m because they are based upon the soil measurements made at the mro for dry conditions reported by @xcite . in cst
we modeled the finite ground plane as a finite sheet in free space .
both simulations resulted in losses @xmath50 1.0% but with opposite frequency dependence trends .
cst results varied between 0.4 and 1.0% , with the larger loss at lower frequencies , while feko varied between 0.2 and 1.0% with larger loss at high frequencies .
the effects are small in both cases and , considering the simulation uncertainties , it is not clear which model is more appropriate .
hence , we took the approximate average of the two simulations as a constant 0.5% loss , indepdendent of frequency . correcting for 0.5% frequency - independent ground loss caused @xmath1 to become more positive by up to 0.003 for 0 h @xmath50 lst @xmath50 12 h and become more positive by 0.003 - 0.006 for 12 h @xmath50 lst @xmath50 24 h as compared to no ground loss correction .
the antenna losses arise from the balun resistance , the connection to the receiver , and the resistance of the antenna panels .
the balun of the edges system includes two brass tubes ( outer diameter 0.5 " ) with a copper - plated brass rod running up one of the tubes and a teflon - dielectric connector to the receiver .
when we account for the impedance in the balun structure and its connection to the receiver in the calibration process ( parameters obtained via transmission line models and lab measurements ) , we see an effect on @xmath1 no greater than 0.015 .
there is a very minor effect on @xmath1 from the panel resistance of 0.001 .
when we use cst and feko to compute the beam correction factor , the difference in the spectral index between methods is 0.016 .
this considers different choices of including or subtracting the cmb as well as different scalings of the haslam 408 mhz map to 150 mhz assuming realistic spectral indices . to assess the impact of spatial variations in the spectral index on our beam correction factor
, we used the frequency dependent gsm instead of the spatially flat scaled haslam map .
the results show that the gsm - derived beam correction factor differs from our default method by 0.3% at most .
the effect upon the spectral index causes a difference of @xmath510.003 for all lst values except from 20 to 23 h where it rises to + 0.005 .
hence , the main component of chromaticity originates from the beam and not spectral index spatial variations .
we now combine the sources of systematic uncertainty .
if we assume our adjustments have a standard error of half of the magnitude of the effect ( with vs. without or method 1 vs method 2 ) , and combine the sources of system error in quadrature : temperature gradients 0.005 , finite ground plane effects 0.006 ; antenna balun 0.015 ; finite panel resistance 0.001 ; and beam correction factor 0.016 , we arrive at a systematic uncertainty on @xmath1 of 0.023/2 = 0.012 . adding the maximum data scatter value of 0.003 directly to the systematic uncertainty ,
as opposed to in quadrature , the total maximum uncertainty in @xmath1 becomes 0.015 .
based on this value we conservatively report a total one - sigma uncertainty of @xmath52 .
this uncertainty improves upon the uncertainties reported in @xcite of @xmath53 considering only edges data and @xmath54 when including haslam skymap data .
additionally , we measure data over a significantly longer time period and include the lst dependence of the spectral index . the final measurement with the full uncertainty range is shown in fig . [ fig : gsm_vs_guz - has ] .
we compare our measured results for the spectral index to simulated results by using publicly available sky maps : the de oliveira - costa gsm ; the haslam 408 mhz map ; and the guzmn 45 mhz map .
we simulate the antenna temperature that edges is expected to observe using the gsm from 90 to 190 mhz in steps of 1 mhz as a function of lst , by convolving the sky model ( less @xmath33 ) with the simulated edges blade beam ( fixed at 150 mhz to compare to the beam - corrected measurements ) .
we then fit the simulated antenna temperature to equation ( [ eq : power_law_1 ] ) to obtain the spectral index vs. lst .
because the haslam and guzmn sky maps are only available at 408 mhz and 45 mhz , respectively , we calculate the spectral index with the closed form expression @xmath55 where @xmath56 is @xmath57 @xmath58 is the achromatic beam for a given pointing and orientation , @xmath39 is frequency , and @xmath59 is the sky model in use .
@xmath58 is normalized to a unit integral at each frequency . in this manner
, we generate spectral index values across 24 hours of lst .
the results of these two simulations along with our result from measured data are shown in fig .
[ fig : gsm_vs_guz - has ] .
we find that the gsm - derived spectral indices tend to over - predict the spectral index of the observations at high - galactic latitudes ( low lst ) .
the gsm - derived spectral indices differ from our measurements by 0.05 < @xmath60 < 0.12 ( 3 to 7.5 sigma ) away from the galactic centre and @xmath61 ( 1 sigma ) near the centre .
the guzmn - haslam ( gh ) derived spectral indices , on the other hand , are a closer match .
the gh indices only differ from our measurements by 0.01 < @xmath6 < 0.05 ( 1 to 3 sigma ) away from the galactic centre and agree well at 18 h lst .
one possible cause for the better agreement to the data by the gh model compared to the gsm is the higher sky temperatures reported in the guzmn map as compared to the gsm at 45 mhz .
away from the galactic centre , the guzmn map is 1100 k higher than the gsm in much of the sky . if the gsm had higher temperatures away from the galactic centre , the simulated spectral index using the gsm maps would decrease for the lower values of lst , giving better agreement with our measurements .
based on these results , we caution that the gsm likely under - predicts the sky temperature at high - galactic latitudes below 200 mhz .
if we make a direct comparison against @xcite where their edges - only spectral index result was @xmath62 at lst = 2.5 h , we report @xmath63 , which is consistent with the 2008 results . to compare against their @xmath64 value , we must use our @xmath29 measurement values at lst = 2.5 h coupled with the @xmath65 values as computed using the haslam sky map with our antenna beam . in this case , we compute @xmath66 as compared to their result of @xmath67 .
the disagreement in these two values arise from differences in @xmath65 and @xmath29 values .
@xmath65 values differ by 0.7 k , which is most likely due to different antenna beam models in use ( due to different antenna designs ) and results in 33 % of the disagreement .
@xmath29 values differ by twice the 2008 stated uncertainty in @xmath29 ( @xmath40 = 10 k ) , which we believe is due to accuracy of the @xcite measurement and accounts for the remainder of the difference .
we measured the sky brightness temperature as a function of frequency and derived the spectral index @xmath1 as a function of sidereal time by fitting to a two parameter model over a span of 240 days using 211 days of nighttime data acquired from july 2015 to march 2016 .
instrument calibration , including corrections for temperature gradients , ground loss , antenna losses , and beam chromaticity , has been applied to deliver instrument stability of over several months as demonstrated by spectral index standard deviation values @xmath68 0.003 .
we have presented results of extensive measurements of the diffuse radio sky between the frequencies of 90 - 190 mhz .
we find that the spectral index @xmath1 is in the range @xmath2 0.02 in the lower lst values , but increases to @xmath43 with the galaxy overhead .
the gsm tends to over - predict the strength of the variation in the spectral index in the range 0.05 < @xmath6 < 0.12 for low lst . however , comparison with the spectral index as computed using the guzmn sky map at 45 mhz and the haslam sky map at 408 mhz is a closer match and differs by 0.01 < @xmath6 < 0.05 away from the centre . at the galactic centre
both models agree with our measurements of the spectral index to within one sigma .
future work is planned to investigate the criteria needed to fit to a three - parameter spectral equation .
updates to sky models will be used to simulate spectral index across lst using the edges blade antenna .
we also plan to measure the spectral index in the @xmath69 mhz range with the low - band edges blade antenna .
the authors would like to thank adam beardsley for useful inputs and discussions .
this work was supported by the nsf through research awards for the experiment to detect the global eor signature ( ast-0905990 and ast-1207761 ) and by nasa through cooperative agreements for the lunar university network for astrophysics ( nna09db30a ) and the nancy grace roman technology fellowship ( nnx12ai17 g ) .
raul monsalve acknowledges support from the nasa ames research center ( nnx16af59 g ) .
edges is located at the murchison radio - astronomy observatory .
we acknowledge the wajarri yamatji people as the traditional owners of the observatory site .
we thank csiro for providing site infrastructure and support .
200 ali z. s. et al . , 2015 , apj , 809 , 61 alvarez , h. , aparici , j. , may , j. , olmos , f. , 1997 , a&a , 124 , 315 bennett c. l. et al . , 2003 , apjs , 148 , 1 bennett c. l. et al . , 2003 , apjs , 148 , 97 bernardi g. , mcquinn m. , greenhill l. j. , 2015 , apj , 799 , 90 bernardi g. et al . , 2016 , arxiv:1606.06006 bowman j. d. , rogers a. e. e. , hewitt j. , 2008 , apj , 676 , 1 bowman j. d. , rogers a. e. e. , 2010 , nature , 468 , 796 bowman j. d. et al . , 2013 pub
australia 30 , e031 bowman j. d. , morales m. f. , hewitt j. , 2009 , apj , 695 , 183 bridle , a. h. , 1967 , mnras , 136 , 219 burns j. et al . , 2012 , advances in space research , 49.3 , 433 davies r. d. , watson r. a. , gutierrez c. m. , 1996 , mnras , 278 , 925 de oliveira - costa a. , tegmark m. , gaensler b. m. , jonas j. , landecker t. l. , reich , p. , 2008 ,
mnras , 388 , 247 fixsen d. j. et al . , 2011 , apj , 734 , 5 furlanetto s. r. , 2006 , mnras , 371 , 867 guzmn , a. e. , may , j. , alvarez , h. , maeda , k. , 2011 , a&a , 525 , a138 haslam c. g. t. , klein u. , salter c. j. , stoffel h. , wilson w. e. , cleary m. n. , cooke d. j. , thomasson p. , 1981
, a&a , 100 , 209 haslam c.g.t . , salter c.j . , stoffel , h. , wilson w.e . , 1982 , astronomy and astrophysics supplement series , 47 , 1 heald , g. h. et al .
, 2015 , a&a , 582 , a123 kogut a. et al . , 2011 ,
apj , 734 , 4 kogut a. , 2012 , apj , 753 , 110 maeda , k. , alvarez , h. , aparici , j. , may , j. , reich , p. , 1999 , a&s , 140 , 145 mellema g. et al . , 2013 ,
astronomy , 36(1 - 2 ) , 235 - 318 .
mesinger a. , ferrara a. , spiegel d , 2013 , mnras , 431 , 621 monsalve r. , rogers a. e. e. , bowman j. d. , mozdzen t. , 2016 , preprint arxiv:1602.08065 mozdzen t. , bowman j. d. , monsalve r. , rogers a. e. e. , 2015 , mnras , 455 , 3890 patra n. , subrahmanyan r. , raghunathan a. , shankar n. u. , 2013 , exp .
astronomy , 36 , 319 platania p. , et al .
, 2003 , a&a , 410 , 847 pober j. , et al . , 2015 , apj , 809 , 62 pritchard j. r. , loeb a. , 2010 , phys .
d , 82 , 3006 remazeilles p. , et al . , 2015 ,
preprint arxiv:1411.3628v2 roberts , 1957 , proc .
ire , 1624 rogers a. e. e. , bowman j. d. , 2008 , aj , 136 , 641 rogers a. e. e. , bowman j. d. , 2012 , radio science , 47.6 rogers a. e. e. , bowman j. d. , vierinen j. , monsalve r. , mozdzen t. , 2015 , radio science , 50 , 130 seiffert m. et al , 2011 , apj , 734 , 6 singh s. , subrahmanyan r. , shankar n. u. , raghunathan a. , 2015 , preprint arxiv:1505.02491v1 sokolowski m. et al . , 2015 , pub . ast .
australia 32 , e004 subrahmanyan , r. , cowsik , r. , 2013 , apj , 776 , 42 sutinjo , a. t. , et al . , 2015 , ieee transactions on antennas and propagation , 63 , 5433 thyagarajan , n. , et al . , 2015 , apj , 804 , 14 tingay s. j. , 2013 , publications of the astronomical society of australia ( pasa ) , vol .
30 , e007 turtle , a. j. , pugh , j. f. , kenderdine , s. , pauliny - toth , i. i. k. , 1962 , mnras , 124 , 297 van haarlem m. p. et al .
, 2013 , a&a , 556 , a2 vedantham h. k. , koopmans l. v. e. , de bruyn a. g. , wijnholds s. j. , ciardi b. , brentjens m. a. , 2014 , mnras , 437 , 1056 voytek t. c. , natarajan a. , garcia j. m. j. , peterson , j. b. , lopez - cruz o. , 2014 , apjl , 782 , l9 wayth , r. b. et al .
, 2015 , pasa , 32 , e025 | we report absolutely calibrated measurements of diffuse radio emission between 90 and 190 mhz from the experiment to detect the global eor signature ( edges ) .
edges employs a wide beam zenith - pointing dipole antenna centred on a declination of @xmath0 .
we measure the sky brightness temperature as a function of frequency averaged over the edges beam from 211 nights of data acquired from july 2015 to march 2016 .
we derive the spectral index , @xmath1 , as a function of local sidereal time ( lst ) and find @xmath20.02 between 0 and 12 h lst .
when the galactic centre is in the sky , the spectral index flattens , reaching @xmath30.02 at 17.7 h. the edges instrument is shown to be very stable throughout the observations with night - to - night reproducibility of @xmath4 < 0.003 . including systematic uncertainty , the overall uncertainty of @xmath1 is 0.02 across all lst bins .
these results improve on the earlier findings of @xcite by reducing the spectral index uncertainty from 0.10 to 0.02 while considering more extensive sources of errors .
we compare our measurements with spectral index simulations derived from the global sky model ( gsm ) of @xcite and with fits between the @xcite 45 mhz and @xcite 408 mhz maps .
we find good agreement at the transit of the galactic centre .
away from transit , the gsm over - predict by 0.05
< @xmath5 < 0.12 , while the 45 - 408 mhz fits over - predict by @xmath6 < 0.05 . [ firstpage ] dark ages , reionisation , first stars - instrumentation : miscellaneous - galaxy : structure |
with the improvement of accelerator technology , various photon sources around the world are set up based on laser compton scattering process @xcite , which possesses several advantages , such as rather flat energy distribution with small spreading and high linear- or circular - polarization @xcite .
the shanghai synchrotron radiation facility ( ssrf ) is designed for the advanced third generation moderate energy synchrotron radiation facility @xcite .
the synchrotron radiation ray produced from the ssrf covers a broad wave band from far infrared to hard x - ray .
it has a low emittance of moderate energy electrons .
the shanghai laser electron gamma source ( slegs ) will be built at one of the straight sections of the ssrf .
gamma - rays with energy up to 22 mev will be produced by compton backscattering of far infrared laser photons on the 3.5 gev electrons circulating in the synchrotron ring .
the slegs will be used for research in astrophysics , nuclear physics and applications for material sciences , etc .
so the proposed construction of slegs will be able to provide us uncommon opportunities to probe deeply sub - nuclear structure and expand our research methods . at present , the slegs is at the stage of construction of its proto type .
the proto type will be set up at the terminal of a 100 mev linac .
x - rays at the energy of 10 kev order can be produced from the proto type . due to the limited capabilities of the 100 mev linac and the laser
, the brightness of x - ray produced from the proto type will be low
. however , this is not so serious , because the primary object for the construction of the proto type is to pave the way for the future setup of the slegs .
the scattering process between the co@xmath0 laser photons and the electrons from 100 mev linac is shown schematically in fig .
[ scattering ] . here
@xmath1 and @xmath2 are the kinetic energies of incident and scattered electron and @xmath3 and @xmath4 are the energies of incident and scattered laser photon in the laboratory frame .
@xmath5 ( if @xmath5 equals @xmath6 , this scattering can be simplified to strict backscattering . ) is the laser incident angle relative to the electron beam direction in the laboratory frame .
after backscattering , the x - ray emerges in the laboratory frame at a small angle @xmath7 relative to the electron beam direction and the electron emerges at the angle @xmath8 .
the theoretical basic of compton backscattering mechanism was described in refs .
@xcite . for a better understanding of this process
, some important relationships will be displayed as below .
one can obtain the energy of scattered photon with the help of equations for energy and momentum conservation : @xmath9 where @xmath10 is the ratio of the electron and light velocities .
considering the electron energies above 50 mev ( @xmath111 ) , and a small scattering angle ( @xmath1210 mrad ) , formula ( 2 ) can be simplified to @xmath13 given by federici et al .
@xcite , sandorifi et al .
@xcite and blumberg @xcite , where @xmath14 is the ratio of the total electron energy and its energy at rest : @xmath15 and @xmath16 is the electron mass at rest .
the differential compton scattering cross section for energetic electrons in the laboratory frame can be obtained starting with its form in the er frame , in which the cross section was derived by klein - nishina @xcite : @xmath17 where @xmath18 , @xmath19 and @xmath20 are the classical electron radius , the ratio between the energies of scattered and incident photon , and the scattering angle in the er frame , respectively .
it can be obtained from eq .
( 2 ) considering the electron to be at rest , i.e. @xmath21 . in this case , the er and laboratory frame are equivalent .
the result is : @xmath22 then one can obtain the compton scattering cross section for energetic electrons in laboratory frame via lorenz transformation : @xmath23 energy differential cross - section of compton backscattering x photons can be derived from formula ( [ energy1 ] ) and ( [ kleinnishina - lab ] ) : @xmath24 when injected laser beam is totally polarized , circular and linear polarization of bcs photons are presented as @xmath25
the proto type of the slegs facility consists mainly of 100 mev linac , co@xmath26 laser optical system , target chamber , detector system and data acquisition system , etc .
they will be introduced in detail in the following . the schematic illustration for the 100 mev linac is shown in fig .
[ linac ] .
it is mainly composed by an electron gun , an pre - buncher and a buncher , four accelerating tubes , a focus coil , a klystron , a modulator , a microwave driving source , a phase control system , a vacuum system , a water - cooling system , a beam flux diagnosis system and a control system , etc .
the total length is about 20 meters .
each accelerating tube may make the electronic energy to increase approximately 27.5 mev .
about 7 mw power from klystron is input to the buncher via the wave guide directional coupler .
the klystron outputs the other microwave power to the four accelerating tubes via the wave guide merit minute , each accelerating tube power accepts about 7 mw . fig . [ optical ] display a schematic view for optical system of proto type . first , the co2 laser will pass through an energy attenuator and then be expanded .
a suitable adjustment of the positions of the two lenses will make the laser light be parallel .
then , the parallel light will be polarized linearly or circularly via the polarizing disc . finally , the laser will be reflected and then be guided to the 100 mev linac hall for the preparation of collision with electron beam . in the 100 mev linac hall
, the laser will be adjusted again , to make it pass vertically through target window .
the details for the adjustment of laser and electron beam will be provide in following . laser optical system . ] in order to reduce the loss of the co@xmath0 laser power , the laser should transmit in a vacuum chamber . according to the present situation of 100 mev linac , we plan to develop a vacuum target chamber , which is used for the entrance of laser . the schematic view for target chamber is displayed in fig .
[ collimation ] .
as well known , the collimation between the co@xmath0 laser and the electron beam is very important , because it decides the energy region and the flux of x - ray .
[ collimation ] .
displays step by step the aim between the co@xmath0 laser and the electron beam .
there are two flakes filmed fluorescence and a mirror with a hole at its center for the transmission of laser .
first , one should initialize the aim via marking the track of electron beam on the two flakes .
second , another laser used for collimation will be imported and adjusted to be coaxial with the electron beam .
then the co@xmath0 laser should be adjust to be coaxial with collimated laser .
the cross marker should be used for the benchmark for the entrance of the co@xmath0 laser .
third , to readjust the co@xmath0 laser after the co@xmath0 laser being fixed at the linac . finally , to readjust the electron beam to make it be coaxial with the co@xmath0 laser . laser and electron beam . ]
the x - ray from compton back scattering performed at the proto type has an energy range of 5 kev@xmath2718 kev and the corresponding wave length range is about @xmath28 cm@xmath27@xmath30 cm ( [email protected] ) .
therefore a hpge detector or a si ( li ) detector is best suitable , because their survey energy scope is about 1 kev@xmath2730 kev .
the efficient area is larger than 12.5 mm@xmath31 and the energy resolution ( fwhm ) is smaller than 160 ev at 5.9 kev .
the related parameters for si ( li ) detector are listed in table [ sili ] .
0.7 cm .basic parameters for si ( li ) detector .
[ cols= " < , < " , ] [ sili ] here the data acquisition refers mainly to the acquisition and processing for data from the detector terminal .
so data acquisition system includes standard camac and the vme data bus system , which are adopted frequently in nuclear physics as well as in high energy physics experiment .
[ daq ] displays the control illustration for daq .
a synch trigger signal is provided by the optical fiber of the ssrf synch system .
then the light signal is transformed by the photoelectricity coupled apparatus into the switch quantity electrical signal , which is used to trigger the co@xmath0 laser .
when the laser output power approaches peak value after an exponential increase with 70 @xmath32s , the co@xmath0 laser synch system will trigger the permission logical gate of the gamma detector .
the detecting time is 30 @xmath32s .
when gamma rays are detected , they will be transformed into electricity pulse , which is proportional to the energy deposits of gamma rays in the detector .
this pulse will be reshaped and enlarged with a maintenance of peak value .
then it passes through zero fixed time to output the a / d transformation logic pulse .
after the a / d converter transforms being finished , a logic signal will be produced to end the maintenance of peak value .
various useful parameters for 100 mev linac and for co@xmath0 laser are listed in table [ parameter ] .
all the required quantities are calculated based on the above formulas and the parameters listed in table [ parameter ] . 0.2 cm l l + electron energy & 100 mev + pulse width & @xmath33324 ns ( fhwm ) + normalized emittance & 107 ( x)@xmath34137 ( y ) @xmath32mrad + energy resolution & @xmath350.9@xmath36 + mean current & @xmath33100 ma + + wave length & 10.6@xmath372 @xmath32 m + peak power & [email protected] kw + rise / fall time & @xmath3590 @xmath32s + stabilization of power & @xmath358@xmath36 + [ parameter ] fig . [ ener - section - pola]a shows the relation between the energy of compton backscattering photons and the scattered angle .
the scattered angles of compton backscattering photons above 10 kev are less than 4.6 mrad which corresponds to a beam size of 3.6 cm at the distance of 4 m from the collision point .
[ ener - section - pola]b displays the differential cross section of compton backscattering photons .
it shows that the differential cross section increases sharply when the scattered angle is less than 5 mrad , which is consistent with interesting energy region above .
the polarization of compton backscattering photons is given in fig .
[ ener - section - pola]c .
it can be seen that if laser light is 100% polarized , a compton backscattered photon is highly polarized at the maximum energy .
the polarization drops as the photon energy decreases as shown in fig .
[ ener - section - pola]c .
however , the energy of laser photons is easily changed , so the polarization can remain be maintained reasonably high in the energy region of interest .
differential luminous flux of compton backscattering photon as a funtion of interaction length at different laser waist widths is shown in the upper part of fig .
[ lumin ] . here
the electron beam is assumed to be zero emittance and its transverse widths are 200 @xmath32 m for @xmath38 direction and 100 @xmath32 m for @xmath39 direction .
these two quantities will be perhaps changed a little in the real operation of the linac .
in addition , the electron beam and the laser are assumed to be co - axis and co - focus .
it is seen obviously that differential flux distributions along interaction length are different at different laser waist widths .
differential flux changes sharply with the transform of interaction position if laser waist width is very small but becomes insensitive when waist width of laser is large enough .
the backgrounds from dipole magnet and bremsstrahlung for the proto type of slegs are estimated and the corresponding results are displayed in figs .
[ synch ] and [ brem ] , respectively .
since the radius of dipole magnet is not decided yet , three candidate radii are assumed in the calculations , which are performed using the program xop 2.1 developed by european synchrotron radiation facility ( esrf ) group .
the fluxes of photons from dipole magnet at three radii are much higher than that of x photons . but
the energies of background are very small ( less than 1 kev ) , which is lower than the energy of x photons of interest .
so the background from dipole magnet will not cover up the x photons of interested energy region .
in conclusion , the design of proto type of slegs at ssrf is introduced .
the related simulation results are presented and discussed .
the backgrounds from dipole magnet and bremsstrahlung are estimated , respectively . from above calculations ,
the ideal obtained flux of x - ray is about 0.076 s@xmath40@xmath41w@xmath40@xmath41nc@xmath40 .
however considering the limited reception of detector and other x - ray loss , the real flux of x - ray will be larger than 0.002 s@xmath40@xmath41w@xmath40@xmath41nc@xmath40 . if a k-500 co@xmath0 laser is selected , its maximum power is about 1500 w. based on the above estimation for background , the signal - noise ratio for the proto type will be larger than 30 .
this work was supported by the innovation program of science @xmath42 technology of chinese academy of sciences under contract no .
kjcx2-sw - n13 , by the national natural science foundation of china under contract no . 10475108 , and by the hundred talent project of sinap .
l. casano , _ et al .
_ , laser unconventional optics j. 55 ( 1975 ) 3 g. matone , _ et al .
notes phys .
62 ( 1977 ) 149 l. federici , _ et al .
_ , nuovo cimento b 59 ( 1980 ) 247 d. babusci , _ et al .
instr . and meth . a 305 ( 1991 ) 19 a. dangelo ,
_ et al . _ ,
instr . and meth . a 455 ( 2000 ) 1 , and references therein .
park , _ et al .
instr . and meth . a 475 ( 2001 ) 425 v.n .
litvinenko , _ et al .
( 1997 ) 4569 h. ohkuma , _ et al .
_ , proceedings of epac 2006 , edinburgh , scotland m. fujiwara , prog . part . and
nucl . phys .
50 ( 2003 ) 487 k. aoki , _ et al .
instr . and meth . a 516 ( 2004 ) 228 d. li ,
_ et al . _ ,
nucl . instr . and meth . a 528 ( 2004 ) 516 e.l .
saldin , _ et al .
instr . and meth . a 362 ( 1995 ) 574 z.t .
zhao and h.j .
xu , proceeding of epac 2004 , lucerne , switzerland a. m. sandorfi , m. j. levine , c. e. thorn , g. giordano , and g. matone , the fabrication of a very high energy polarized gamma ray beam facility and a program of medium energy physics research at the national synchrotron light source , brookhaven national laboratory report bnl-32717 , bnl physics dept .
proposal to the dept . of energy ( 1982 ) .
l. n. blumberg , kinematics of compton backscattering x - ray source for angiography , brookhaven national laboratory report bnl-47503 ( 1992 ) . | the design for the proto type of the shanghai laser electron gamma source ( slegs ) at the shanghai synchrotron radiation facility ( ssrf ) is introduced .
some detailed descriptions for design of related instruments are provided .
the proto type can produce x - ray with energy of 10 kev order .
a description of the kinematics of compton backscattering mechanism and the related simulation results are presented and discussed .
the backgrounds from dipole magnet and bremsstrahlung are estimated and the signal - noise ratio is also given .
, , , , , , , , , , , , , , compton backscattering , 100 mev linac , co@xmath0 laser , x - ray 13.60.fz , 41.60.ap , 41.75.ht , 42.55.lt , 07.85.fv |
the operation of any computing device is necessarily a physical process , and this fundamentally determines the possibilities and limitations of the computing machine .
a common thread in the history of computers is the exploitation and manipulation of different natural phenomena to obtain newer forms of computing paradigms @xcite .
for instance , chaos computing @xcite , neurobiologically inspired computing , quantum computing@xcite , and dna computing@xcite all aim to utilize , at the basic level , some of the computational capabilities inherent in natural systems .
in particular , larger understanding of biological systems has triggered the interesting question : what new directions do bio - systems offer for understanding and implementing computations ?
the broad idea then , is to create machines that benefit from natural phenomena and utilize patterns inherent in systems to encode inputs and subsequently obtain a desired output .
different physical principles can yield logic outputs , for logic gates such as and , or , xor , nor and nand which form the basis of the universal general - purpose computation .
it is particularly interesting if systems of biological relevance can also yield logic outputs consistent with the truth tables of different logic functions @xcite . in this work
we will explore the computational possibilities arising from the dynamics of a genetic network with quorum sensing feedback .
quorum sensing allows cells to sense their own density and to communicate with each other , giving the cells the ability to behave as a population instead of individually .
the autoinduction of luminescence in the symbiotic marine bacterium _
vibrio fischeri _ is perhaps the best characterized example of quorum sensing .
some other examples include sporulation and fruiting body formation by _
myxococcus xanthus _ and antibiotic production by several species of _ streptomyces_. the way quorum sensing is accomplished in cells is through diffusive exchange of auto - inducer molecules which participate in intercellular coupling and self feedback . at low cell density ,
the auto inducer is synthesized at basal levels and diffuses into the extracellular space and gets diluted .
but as the cell density increases , the intracellular autoinducer concentration increases until it reaches a threshold beyond which it is produced auto - catalytically .
this results in a dramatic increase of product concentrations , thus creating bi - stability between a high steady state and a low steady state @xcite .
specifically , in this work we focus on an artificial genetic network known as the repressilator , with feedback of the kind that quorum sensing may provide to an isolated repressilator .
a repressilator consists of a ring of three genes , products of which inhibit each other in cyclic order .
the quorum sensing feedback stimulates the activity of a chosen gene providing competition between the inbitory and the stimulatory activities localized in that gene . in this work
we attempt to use the varying concentrations of proteins in the repressilator , arising due to quorum sensing feedback , to realize different logic gates . in the sections below we
will first discuss the model of the repressilator genetic network with quorum sensing feedback , and present the change in dynamics of repressilator proteins with changing parameters of the quorum sensing feedback .
we will then go on to use this dynamics to implement a range of logic functions , by appropriate input - output associations .
we consider a basic genetic ring oscillator network in which three genes inhibit each other in unidirectional manner , and an additional quorum sensing feedback loop stimulates the activity of a chosen gene , providing competition between inhibitory and stimulatory activities localized in that gene . this system is known to yield behaviour ranging from limit cycles , to high and low stable steady states @xcite . in the model , 3 genes in a cyclic loop forms mrna ( @xmath0 ) and proteins ( @xmath1 ) .
the three genes inhibit each other , where one gene is inhibited by the preceding gene .
quorum sensing feedback acts between gene @xmath2 and @xmath3 , where @xmath2 stimulates the production of @xmath3 by quorum sensing along with an inhibitory action .
the model can be reduced to fast mrna kinetics , where @xmath0 is assumed to be in a steady state with their respective proteins ( @xmath1 ) , leading to @xmath4 .
the equations describing the dynamics of this system then are : @xmath5 here @xmath6 denotes the concentration of the auto - inducer ai .
the relevant parameters in this system are the ratio of protein decay rate to mrna decay rate @xmath7 , maximum transcription rate in the absence of an inhibitor @xmath8 , hill co - efficient for inhibition @xmath9 , diffusion co - efficient depending on the permeability of the membrane to auto - inducer molecule @xmath10 , the rate of auto - inducer decay to mrna decay rate @xmath11 , rate of auto - inducer production @xmath12 , and most importantly , the co - efficient for the production of @xmath3 protein due to auto - inducer @xmath13 and the concentration of auto - inducer in external medium @xmath14 responsible for the _ quorum sensing feedback_. in our simulations we fix the parameters at realistic values @xcite : @xmath15 , @xmath16 ; @xmath17 ; @xmath18 , and @xmath19 ( relevant to slow protein kinetics ) .
we vary the parameters that determine the strength of quorum sensing , namely @xmath14 and @xmath13 . on taking different values of parameters @xmath14 and @xmath13 , the strength of quorum sensing changes , and hence the qualitative nature of the dynamics changes .
this is clearly evident from the bifurcation analysis of the concentrations of protein @xmath20 and @xmath2 with respect to parameters @xmath14 and @xmath13 displayed in figs .
[ bifsext]-[bifkappa ] , which shows the change in dynamics from limt cycles to stable steady states as the external concentration of auto - inducer increases .
there is also a sudden switch in the value of the steady state after a critical value of @xmath14 and @xmath13 .
now , in order to utilize this system to realize a logic gate we need to propose an appropriate _ input - output mapping _ , namely , we must first suggest how inputs and outputs are encoded .
the task at hand then , is to ensure that this input and output encoding _ consistently yields the correct input - output association corresponding to different logic functions _ , as displayed in the truth table ( table i ) .
specifically , here we will implement a series of logic gates , such as and / nand , or / nor and xor / xnor , using the change in dynamics of protein @xmath20 and @xmath2 .
further we will obtain logic operations in parallel by simultaneous measurement of the concentrations of these two proteins .
+ * encoding logic inputs : * + consider the inputs to be encoded by @xmath14 , namely the concentration of auto - inducer in external medium , which is responsible for the quorum sensing feedback . we propose that the two inputs @xmath21 where @xmath22 and @xmath23 encode the two logic inputs as follows : a. when logic input is @xmath24 , @xmath25 ; b. when logic input is @xmath26 , @xmath27 ; so the pair of inputs lead to the following three distinct conditions : a. @xmath28 for input set @xmath29 b. @xmath30 for input set @xmath31 c. @xmath32 for input set @xmath33 + * encoding logic output :
* + we consider the _ maximum protein concentration as an indicator of the output of the system_. specifically we consider the maximum concentrations of proteins @xmath20 and @xmath2 , denoted by @xmath34 and @xmath35 respectively .
this gives us the ability to obtain two outputs from the system in parallel .
now a prescribed output determination threshold , @xmath36 and @xmath37 , leads to the mapping of the concentration to a @xmath38 logic output , as folllows : for protein @xmath20 : a. if @xmath39 , then the logic output is @xmath24 b. if @xmath40 , then the logic output is @xmath26 and for protein @xmath2 : a. if @xmath41 , then the logic output is @xmath24 b. if @xmath42 , then the logic output is @xmath26 one can obtain the _
complementary gates _ by exchanging the output determination criterion .
namely : for protein @xmath20 : a. if @xmath43 , then the logic output is @xmath24 b. if @xmath44 , then the logic output is @xmath26 and for protein @xmath2 : a. if @xmath42 , then the logic output is @xmath24 b. if @xmath41 , then the logic output is @xmath26 + * controlling the nature of the logic function * : + we can control the type of logic operation obtained by simply changing the output determination threshold .
namely , we can obtain and / nand , or / nor or xor / xnor input - output mappings by considering different @xmath36 and @xmath37 .
for instance , for realizing the fundamental nand logic gate ( see truth table i ) , we have the binary logic output determined by the threshold level @xmath45 .
namely , if the maximum concentration of protein @xmath20 , @xmath34 is greater than @xmath36 , then the logic output is @xmath26 , and if @xmath46 , then logic output is @xmath24 .
this is clearly evident in fig .
[ nand_a ] and table [ nand_a_table ] . in order to change the logic function from the fundamental nand gate to the funademental nor gate
, we simply have to change the prescribed output determination threshold .
specifically , the system yields nor logic when the output determination threshold @xmath36 is @xmath47 . similarly , using different output determination thresholds @xmath37 for protein concentration @xmath2 , we can obtain and and xnor logic operations .
this is clearly evident through fig .
[ and_b ] and table [ and_b_table ] .
again note that the complementary logic functions nand and xor are implemented by a simple toggle of the logic output determination condition .
so we can obtain any combination of and / nand / or / nor in parallel with and / nand / xor / xnor logic through the two protein concentrations for the same pair of inputs . in fig .
[ timeseries ] , the time series plots for protein @xmath20 and @xmath2 exhibits the simultaneous realization of nand and xnor gates respectively , for a representative random stream of input values .
clearly , this same system can yield xor and and in parallel by the alternate output association , namely a logic output @xmath26 is obtained when the proteins @xmath20 and @xmath2 are below output threshold @xmath36 and @xmath37 respectively , and @xmath24 otherwise .
in fact , the combination of and and xor in parallel is particularly useful , as it forms the basis of the ubiquitous _ bit - by - bit addition_. alternately , the logic operation may also be changed by a constant shift in input encoding , namely @xmath48 , where different values of @xmath49 yield different logic functions .
for instance , using protein @xmath2 for output determination , and keeping level @xmath37 fixed at @xmath50 , @xmath51 yields xnor , and @xmath52 yields nand .
similarly , using protein @xmath20 for output determination , with @xmath53 , @xmath51 yields nand , and @xmath54 yields nor .
+ * using activation rate parameter @xmath13 to encode logic inputs * : + we now consider an alternate input encoding , with the logic inputs determining @xmath13 . here
@xmath14 is held fixed at @xmath55 , and varying @xmath13 yields different protein concentrations , which leads to different outputs ( see figs .
[ kappa_a]-[kappa_b ] ) .
specifically , let @xmath13 be determined by the logic inputs @xmath22 and @xmath23 as follows : @xmath56 with @xmath57 when logic input is @xmath24 , and @xmath58 when logic input is @xmath26 .
taking a base value of @xmath59 , we have : a. @xmath60 when input set ( @xmath61 ) is @xmath62 b. @xmath63 when input set ( @xmath61 ) is @xmath64 c. @xmath65 when input set ( @xmath61 ) is @xmath66 the binary logic output is obtained from the maximum concentration of proteins @xmath20 and @xmath2 , as before .
namely , we determine the logic output through a threshold : if @xmath67 , then logic output is @xmath24 , and if @xmath67 , then logic output is @xmath26 . as before ,
the nature of the logic function obtained can be controlled by the output determination threshold for the proteins @xmath20 and @xmath2 .
when @xmath68 we obtain nand logic and for @xmath69 we obtain the fundamental nor operation .
similarly , when @xmath70 we obtain and and for @xmath71 we obtain xnor .
representative values are displayed in tables [ kappa_a_table ] and [ kappa_b_table ] . hence , xnor / xor / and / nand and nand / and / nor / or gates can be realized in parallel for the same set of inputs .
namely , by setting the appropriate output determination condition , the protein concentrations can yield any of the logic functions xnor / xor / and / nand and nand / and / nor / or independently and simultaneously .
in summary , we have demonstrated how a genetic ring oscillator network with quorum sensing feedback can operate as a robust logic gate .
we use the concentration of auto - inducer in external medium @xmath14 , and alternatively an activation rate parameter @xmath13 ( cf .
1 ) , to encode logic inputs .
both these quantities regulate the strength of the quorum sensing feedback .
the concentrations of proteins @xmath20 and @xmath2 determine the logic outputs .
this input - output association yields a range of logic functions , namely and / nand , or / nor and xor / xnor .
so the system can act as a versatile unit that can morph between different logic operations .
we further demonstrated the capacity of this system to yield different logic operations simultaneously through the two proteins @xmath20 and @xmath2 .
these observations may provide an understanding of the computational capacity of systems with quorum sensing feedback .
it also may have potential relevance in the design of biological gates , with the ability to flexibly reconfigure logic operations , and implement logic operations in parallel .
further , since electronic analogs of such systems have already been implemented @xcite , our results may readily lead to bio - inspired computing devices .
+ 1 j. p. crutchfield , w. l. ditto , and s. sinha , introduction to focus issue : intrinsic and designed computation : information processing in dynamical systems beyond the digital hegemony , _ chaos _ * 20 * 037101 ( 2010 ) s. sinha and w.l .
ditto , _ phys .
* 81 * ( 1998 ) 2156 ; k. murali , s. sinha , w.l .
ditto , a.r .
bulsara , _ phys .
* 102 * ( 2009 ) 104101 ; w. l. ditto , a. miliotis , k. murali , s. sinha , and m. l. spano , _ chaos _ * 20 * ( 2010 ) 037107 s. james , p. nilsson , g. james , s. kjelleberg and t. fagerstrm ( 2000 ) .
luminescence control in the marine bacterium vibrio scheri : an analysis of the dynamics of lux regulation .
296 , 1127 - 1137 .
garcia - ojalvo , j. , elowitz , m.b . and strogatz , s.h .
: modeling a synthetic multicellular clock : repressilators coupled by quorum sensing .
proceedings of the national academy of sciences of the u.s.a .
101 , 10955 - 10960 ( 2004 )
. .truth table of the basic logic operations for a pair of inputs : @xmath61 @xcite .
note that nand and nor are fundamental logic gates from which any logic circuit can be constructed @xcite . here and and nand , or and nor , xor and xnor are complementary pairs and are simply given by the not operation on the logic output , where not gives value @xmath26 when input is @xmath24 and @xmath24 when input is @xmath26 .
+ [ cols="^,^,^,^,^,^,^",options="header " , ] | we demonstrate how a genetic ring oscillator network with quorum sensing feedback can operate as a robust logic gate .
specifically we show how a range of logic functions , namely and / nand , or / nor and xor / xnor , can be realized by the system , thus yielding a versatile unit that can morph between different logic operations .
we further demonstrate the capacity of this system to yield complementary logic operations in parallel .
our results then indicate the computing potential of this biological system , and may lead to bio - inspired computing devices . |
diquark condensation in the strong interaction ground states at asymptotically high baryon density is a very like possibility @xcite since the dominant one gluon exchange interaction between quarks in such a situation is attractive in the color triplet channel which causes the bcs instability @xcite at the fermi surface for the quarks . at zero and low densities compared to the nuclear saturation
one , the study of the phenomenon in qcd , now called _ color superconductivity _ , is more difficult due to the fact that one has to deal with non - perturbative and finite density effects .
it is quite apparent that the strong interaction vacuum is not color superconducting at the present day condition in a large and uniform region in space since color is known to be confined .
the following question can nevertheless be asked : _ how far away ( in energy density ) is certain type of metastable color superconducting phase , called a virtual phase , of the hadronic vacuum from the stable one at the present day condition?_. it is speculated in refs .
@xcite that certain kind of color superconducting phase exists , which has an energy density close to that of the normal phase for the strong interaction vacuum in which the chiral symmetry is spontaneously broken down .
whether of not such a speculation reflects reality is not a question that can be easily answered theoretically based on qcd since unlike at high densities , the qcd lagrangian at low density is dominated by non - perturbative effects .
lattice study of qcd at finite density is still facing difficulties .
the properties of color superconducting phases inside a nuclear matter with low or intermediate baryonic density was investigated in the literature based on four fermion interaction models @xcite , which start in the early 90s and on instanton motivated models @xcite recently , which is currently actively studied .
however model studies will not be able to determine with confidence when does certain kind of color superconductivity will appear as the density increases .
the question can be answered using experimental means at high energy due to the fact that for high energy processes , which probe the small distance properties of the hadronic system , even a close by metastable vacuum phase can contribute due to non - perturbative quantum fluctuations that sample the contributions from the quasiparticles of the metastable phase of the vacuum @xcite .
direct detection of the virtual phase of the vacuum on the vacuum state itself is very difficult if possible at all .
certain finite hadronic system is needed to serve as a medium in order for significant effects of the possible color superconducting virtual phase of the strong interaction vacuum state to manifest .
one of the best hadronic systems for that purpose is a single nucleon . besides the theoretical simplicity of a nucleon ,
this is also mainly due to the fact that a nucleon much better known experimentally compared to a heavy nucleus so that anormalies with less uncertainties can be found . on the contrary , our combined theoretical and experimental knowledge about a heavy nucleus a lot fuzzy for our purpose .
the effects of the possible virtual color superconducting phase of the hadronic vacuum state can be observed as long as the virtual phase is close enough to the stable normal phase for hadronic systems at zero density and temperature .
this is due to the fact that the concentration of matter / energy inside of a nucleon generate significant signal which is going to be discussed in detailed in the sequel .
a nucleon in either of the above mentioned situations is called a `` superconducting nucleon '' in the rest of the paper .
it should not be confused with the naive pictures , which require that there is in fact a true color superconducting phase or there are certain kind of bound state of diquark or quark quark clustering inside of a nucleon , which , despite of the fact that are frequently used , may be disfavored theoretically @xcite .
albeit this later extreme pictures are logically covered by this paper they are not neccessary ones ( see also @xcite ) .
work in the above mentioned direction had been done in ref .
@xcite , which lead us to belief that there could be in fact a close by virtual color superconducting phase for the strong interaction vacuum .
this work provides a more systematic and model independent study of the possibility .
since color is confined in nature at the present day condition , it is extremely difficult , if possible at all , to measure the color conductivity induced phenomenon of the hadronic systems to make a direct assessment concerning whether or not a color superconducting phase is contributing or not in a given hadronic system . on the other hand , the electromagnetic ( em ) superconductivity of a system can be probed using electrically charged particles like charged leptons and hadrons since the electromagnetic charge is not confined .
an earlier study were carried out @xcite concerning the high energy em properties of a nucleon if it is superconducting in the above mentioned sense .
the investigation is continued into the deep inelastic scattering ( dis ) processes of leptons and nucleons here by trying to explain , in an essentially model independent way , a collection of old and new anomalies about the nucleon in a logically coherent way .
these anomalies are about the peculiar behavior of the elusive `` pomeron '' in hadron
hadron collision and in dis at small bjorken scaling variable @xmath0 and in certain exclusive processes .
the purpose of this work is not to contradict the current explanations of some these problems in perturbative qcd .
it tries to provide a complementary view , based on regge theory for strong interaction , about the physical processes in the intersection area of soft and hard scattering kinematic region . since this work , which contains no detailed approximate computational scheme and model information , only provides constraints due to symmetry considerations that any work which properly implements the symmetries considered is expected to arrive at basically the same results .
the small @xmath0 region of the nucleon structure function , where the perturbative qcd calculation is no longer strictly valid , is an intersection region in which both the perturbative and the non - perturbative phenomena play their role .
we do not understand the non - perturbative qcd well enough from a first principle point of view due to the lack of an effective and practical scheme to tackle the problem .
we do have , however , a phenomenologically successful one based on regge asymptotics , which determines the power law behavior of the hadronic scattering amplitude in the large energy limits that tells us something about the small and large @xmath0 behavior of the structure function . due to the constraint imposed by unitarity , the physical hadronic amplitudes are bounded from above by the froissart bound which requires that the total hadron hadron scattering cross section @xmath1 with @xmath2 the total energy .
the reggeon exchanged that is responsible for a behavior like @xmath3 is called the soft pomeron with an effective intercept of @xmath4 . a high energy virtual photon interacts with the nucleon through a coupling to the charged quarks inside the nucleon .
since quarks are the basic building block of the nucleon , it is quite natural to assume that the virtual photon and nucleon compton scattering amplitude respect the hadronic froissart bound too , provided that the possible final state phase space of these two reactions is of the same nature . the recent experimental observation in lepton deep inelastic scattering ( dis ) and drell yan processes involving a nucleon made it possible to extract the unpolarized as well as polarized structure functions @xmath5 @xcite and @xmath6 @xcite respectively at small @xmath0 with large momentum transfer @xmath7 , the flavor unsymmetry of the nucleon sea quark distribution @xcite and the charge symmetry breaking strength @xcite .
the observed rapid rise of the structure functions in the small @xmath0 region was unexpected from a straight forward extrapolation of the perturbative qcd pictures @xcite .
it turns out that the soft pomeron can not provide the rapid rise of @xmath5 at small @xmath0 @xcite either . instead ,
for a @xmath8 scattering , there is a transition region in the photon energy of order @xmath9 after which the so called hard pomeron with an intercept of @xmath10 seems to be needed in order the reproduce the data @xcite .
such a behavior can not continue all the way down to @xmath11 if the hard pomeron is of hadronic nature since it would lead to a violation of froissart bound due to unitarity .
therefore although two gluon model of the hard pomeron based on the qcd evolutoin equations is capable of giving the proper rise of gluon density at small @xmath0 @xcite , the above mentioned unitarity problem remains in such an approach .
although it can be argued that the true strong interaction asymptotics is still far from our current accessible energy scale , the physics dictated by the qcd bfkl evolution equation may not be the whole story for the small @xmath0 physics for reasons not related to its apparent violation of the froissard bound @xcite .
the observation that the center line of the flavor asymmetry can not fully account for the violation of the gottfried sum rule @xcite means that there seems to be less than three valence quarks inside a nucleon @xcite when interpretated in a straight forward way . where is the missing quark number ?
the charge symmetry breaking in the nucleon structure function obtained in refs .
@xcite also needs to be explained .
the nucleon structure function @xmath5 extracted from high precision neutrino dis on a nucleon @xcite disagree with the one extracted from the charged lepton dis on the same nucleon at small @xmath0 ( @xmath12 ) .
it is still not clear what is the origin of this difference , which could be caused by experimental systematic errors or it could be of physical in nature as anticipated in ref .
if the later possibility is true , then how can such a behavior be understood ?
in a recent analysis of experimental data , the vector current conservation is in question @xcite .
can such a situation be incorporated into the normal picture of hadronic system without certain drastic changes to our faith to symmetries ? why does the polarized structure function seems to change so rapidly which is reported in ref . @xcite and implied in ref .
@xcite at small @xmath0 that no theory @xcite seems can explain its origin at the present ? admittedly , any individual piece of experimental information available at the present may not be sufficiently accurate to lead to a final conclusion , a coherent theoretical study of a collection of them is expected to provide at least a stronger motivation for us to deepen our understanding of the problem both theoretically and experimentally .
the physical processes behind the spontaneous partial breaking of the em local @xmath13 gauge symmetry is discussed in section [ sec : pblgs ] .
the manifestation of the possible spontaneous partial breaking of the em local gauge symmetry at not so high density in the high energy semi - leptonic processes of hadronic systems is studied in section [ sec : slp ] .
the small @xmath14 region of the high energy semi - leptonic scattering is discussed in section [ sec : q20 ] .
the dis region is discussed in section [ sec : dis ] .
an explanation of some of the current puzzles concerning the nucleon in the dis processes is provided .
section [ sec : summary ] contains a discussion and a summary .
the elucidation of the physical processes behind the spontaneous breaking of fundamental local gauge symmetries provides an important step for not only the construction of fundamental theories of nature , like the successful standard model of electroweak interaction , but also the further understanding of the physical properties of the superconducting phase of certain condensed matter systems .
the historical development is covered in ref .
the discussion of the spontaneous breaking of the em gauge symmetry and that of the color @xmath15 gauge symmetry in the color superconducting phase along the traditional lines are given , e.g. , in refs .
however the current understanding of such a process , as it is briefly described in the following subsection , stops at the situation in which the whole charge for the local symmetry involved are spontaneously broken . with the deepening of our knowledge about the structure of matter , especially the hadronic systems ,
new situations emerge in which the above mentioned mechanism becomes no longer suitable .
they arise because the charges for the corresponding fundamental local symmetries can be decomposed as a superposition of various components that generate certain global symmetries of the system .
a familiar example is the electric charge of a fundamental particle like a quark or a lepton .
it can be written generically as @xmath16 in standard model , where @xmath17 is the baryon number contribution to the total electric charge , @xmath18 is the lepton number contribution and @xmath19 is the isospin charge contribution , etc .
it is possible that the ground or vacuum state of the system does not break the whole charge @xmath20 but only its certain component which generates global rather than local symmetry transformations , like part of the baryonic charge @xmath17 of the system .
in such a case @xcite , the electromagnetic gauge symmetry is said to be _ spontaneously partial broken_. but before embarking on understanding the physical processes behind the spontaneous partial breaking of a local gauge symmetry , let us briefly review the relatively well known higgs mechanism for the spontaneous full breaking of a gauge symmetry .
the gauge transformation in a system possessing the corresponding gauge symmetry can be classified into two categories 1 ) the gauge transformation of the first kind and 2 ) that of the second kind .
consider the simplest case of @xmath13 locally symmetric gauge theories .
the symmetry transformations of the first kind correspond to global ones generated by the charge operator @xmath21 of the system ; the transformations of the second kind corresponding to local realization of the global ones are generated by the gauss operator @xmath22 where @xmath23 is the charge density operator and @xmath24 is the `` electric '' field operator .
the existence of a local gauge symmetry of the theory is characterized by the existence of a superselection sector in the hilbert space of the system , which is called the space of physical states @xmath25 , within which the matrix elements of @xmath26 vanish , namely @xmath27 due to the local gauge symmetry of the system , the superselection sector @xmath28 remains invariant during the time evolution of the system-no transition from and to the other sectors of the hilbert space @xmath29 , which are called the subspace of unphysical states , is present . in case of the gauge symmetry of the system
is spontaneously broken down , the vacuum state is not annihilated by the charge operator @xmath30 , namely @xmath31 leading to massless goldstone boson excitation of the symmetry breaking according to the goldstone theorem .
these goldstone bosons do not belong to the physical superselection sector defined by eq .
[ phys - state1 ] of the corresponding local gauge theory so that the otherwise long range force generated by the goldstone bosons corresponding to a spontaneous breaking of the global symmetry of the first kind are actually absent in physical processes due to the existence of the corresponding symmetry of the second kind generated by eq .
[ u1q2 ] . from the definition of the physical states
, it can be shown that the goldstone bosons , generated by acting upon the vacuum state by the total charge operator , belong to the subspace of unphysical states @xmath29 . therefore the massless goldstone bosons associated with the spontaneous breaking of the symmetry transformations of the first kind are absent in the physical processes due to the gauge symmetry .
the process of decoupling of the world - be goldstone bosons is often called the higgs mechanism in high energy physics . this mechanism can be presented in a way different from the more familiar text - book discussion of the higgs mechanism , which is based upon a lagrangian with higgs fields appearing at the tree - level .
we are interested not only in such a possibility , but also the possibility that the `` higgs '' particles are not elementary ones but composite excitations of the system that are absent at the tree - level .
for that purpose , a vertex functional representation is more appropriate .
as it is known , if a symmetry of the first kind is spontaneously broken , then , the corresponding goldstone bosons appear in the vector current vertex of fermions , which is required by the ward takahashi identity relating the divergence of the vector current to the self - energy of the fermions in the system @xcite . in this case , the charge of the fermion is only partially concentrated on the fermionic ( quasi-)particle , which forms a charge core , the rest of the charge is spreaded around it due to the existence of the massless world - be goldstone boson .
this is shown graphically in fig .
[ fig : charge - spread ] and expressed in the following in terms of current operator @xmath32 where the core current is denoted as @xmath33 and the spreaded piece of it , which has a strength characterized by a scalar function @xmath34 , is longitudinal with a massless pole in @xmath14 .
the relative strength between the vector current @xmath35 and the scalar one @xmath34 is determined by the current conservation , namely , @xmath36 the ward
takahashi identity , which ensures the conservation of charge , relates the self - energy and @xmath34 @xcite .
it has the form @xmath37 = \left ( \begin{array}{cc } 0 & ed \\ -ed & 0 \end{array } \right ) \label{jsprd - sig}\ ] ] in the 8-component real " representation for fermions with @xmath38 the basic charge unit , @xmath39 the self - energy of the fermions , @xmath40 the third pauli matrix and the right hand side of the above equation acting on the upper and lower four components of the 8-components fermion spinor .
the quantity @xmath41 is related to the order parameter for the symmetry breaking . for a more concrete discussion , let us consider the case of scalar fermion pair condensation studied in refs .
@xcite . define the goldstone boson fermion coupling vertex as @xmath42 with @xmath43 given by @xmath44 and @xmath45 the order parameter ( see refs .
@xcite ) . then using the graphical representation of the spreaded component of the current given in fig .
[ fig : jspread ] , one finds the value of the goldstone boson fermion coupling constant defined above to be @xmath46 with @xmath47 defined in the following loop diagram @xcite @xmath48 since the photon couples to the vector current of the fermions , the massless pole in the vector current vertex of the fermions in the symmetry breaking phase of the system modifies the photon behavior drastically .
the fully dressed photon propagator @xmath49 includes the self - energy insertion , namely , @xmath50 in a diagrammatic representation with @xmath51 the proper self - energy of the photon and @xmath52 the photon propagator in free space in the gauge such that @xmath53 . in the phase of the vacuum in which the em @xmath13 gauge symmetry is spontaneously broken , the scalar quantity @xmath54 acquires a single pole in @xmath14 , namely , @xmath55 with @xmath56 depending on the order parameter of the symmetry breaking phase . here
@xmath57 is the fine structure constant .
the dyson equation for the full photon propagator can be easily solved in the lorentz gauge . keeping only the pole term of @xmath54 , the photon propagator has the following generic form in an infinite system @xcite
@xmath58 the fermion
fermion scattering in the vector channel represented by fig .
[ fig : ffscattering ] is mediated by the exchange of the media modified photon with its propagator given by eq .
[ photon - prop ] . from eqs .
[ charge - sprd ] and [ photon - prop ] , it follows that only the core part of the current density @xmath35 participates in the scattering .
namely , the scattering amplitude @xmath59 is given by @xmath60 which is generated by `` less '' charge ( compared to that of the original free fermions ) .
therefore the scattering strength between fermions is modified by the symmetry breaking .
the spread charge current density @xmath34 also contribute to the fermion
fermion scattering ; it is mediated by an exchange of the massless goldstone boson .
using the @xmath13 ward
takahashi identity for em @xcite , it can be shown that the goldstone boson contribution to the fermion
fermion scattering is of the form @xmath61 with @xmath62 which is represented in fig . [
fig : ffscattering].b . here
[ gammas ] , [ jsprd - jcore ] , [ gvalue ] and [ photo - mass ] are used .
therefore , the total scattering amplitude is @xmath63 , which is of the form @xmath64 with the full propagator @xmath65 given by @xmath66 the effective full propagator @xmath65 for the interaction is the standard one for a massive vector particle with mass @xmath67 . before proceeding to the next subsection ,
let us make a few remarks : 1 ) although the core current density is used in the scattering amplitude @xmath59 , which weakens the interaction vertex , the particle exchanged has an additional longitudinal component .
it is in this way that the gauge symmetry is maintained 2 ) the massless pole corresponding to the goldstone boson and the massless pole of the photon disappear in the final physical scattering amplitudes .
therefore the originally non - existent goldstone boson in the lagrangian do not appear in the final results either ; it only appear in the intermediate steps of the discussion .
so it does not belong to the subspace physical states .
what exchanged in the fermion
fermion scattering is a massive vector boson with standard propagator eq .
[ pfull - prop ] for _ any charged particles inside of the system _ 3 ) it is evident from eq .
[ tldef ] that the coupling constant @xmath68 between the goldstone boson is combined with the original massless photon to generate an effective massive vector excitation . despite that the goldstone boson couples to the fermions with a strength @xmath68 which is of order 1 due to the fact that the spontaneous partial breaking of the em gauge symmetry is caused by strong interaction , this massive vector particle only couples to the fermions with a strength of em interaction due to a delicate cancellation of its strong interaction effects .
the charge operator in eq .
[ u1q1 ] contains only one piece since the system understudy contains only one charge .
gauge symmetry in the color superconducting phase of quarks system interested in this work also belong to this category ! ] the electric charge operator in the realm of elementary particle physics contains multiple pieces . in the standard model
, the electric charge operator can be decomposed like in eq .
[ charge - decomp ] .
in addition , the baryon number contribution @xmath17 can be further decomposed into components associated with each of the three generations of hadrons , namely @xmath69 with @xmath70 the contributions of the first generation of quarks consists of the light up ( u ) and down ( d ) quarks to the baryon number .
the electric charge density operator can be decomposed in the same way .
although the symmetry generated by @xmath71 has an associated local gauged one with a superselection sector of physical states , its components do not always has a local symmetry attached to them .
if there is a spontaneous breaking down of a ( global ) symmetry corresponding to any one component of the electric charge , then the associated goldstone bosons do not necessarily belongs to the subspace of unphysical states ( for the em gauge symmetry ) and they are expected to participate in the physical scattering processes in certain reaction channels .
the goldstone bosons still can has certain overlap with the physical superselection sector for the em gauge symmetry in the hilbert space of the system after the unphysical component is projected out using eq.[phys - state1 ] .
as it is discussed above , the goldstone boson for a single charge system is not a physical excitation .
the effective excitation that mediates the interaction between the charged particles is a massive vector boson .
the massless excitations decouple from the physical spectra .
these occur due to a delicate cancellation of the long range interaction effects between the longitudinal components of @xmath49 and @xmath72 that couples to the emitter and receiver with a strength in the range of the _ strong interaction _ rather than that of the em interaction .
such a cancellation is protected by the gauge invariance against higher order corrections . for a multi - charged system like the standard model ,
the situation becomes a little complicated . for the scattering between two particles with the same broken charge ( e.g. , the nucleon charge in the light quark system ) ,
the cancellation between the long range interaction between @xmath49 and @xmath72 still occur @xcite .
therefore a particle experiences a short ranged em interaction when meets a particle with the same broken charge in the superconducting phase . in this case , the higgs mechanism is functioning .
what happens if two particles , a and b , with different broken charge meet ?
assuming that particle a radiates a photon with a strength determined by its own electric charge and a goldstone boson with a strength determined by its broken charge ] .
the photon will propagate with medium modified propagator @xmath49 on its way to particle b and the goldstone boson will propagate with propagator @xmath72 on its way also to particle b. particle b receives the photon with the response determined by its electric charge and the goldstone boson with the response determined by its own broken charge .
cancellation between the two forces induced by the longitudinal component of the medium modified em field and that of the goldstone boson does not happen .
the residue force couples particle a and b with a strength of that of strong interaction since the goldstone boson couples a and b in such a way .
the same phenomenon happens if the role of a and b is exchanged .
this can be putted in another way . in a multi - charged system in its superconducting phase which breaks the em gauge symmetry , the _ effective
_ em field strength produced by a charge particle is _
observer dependent_. for a particle with the same broken charge as the source , it sees , effectively , a field of massive ( short ranged in the static limit ) vector particle with an strength of em interaction . for a particle with different broken charge , what it effectively sees is two force fields : the first one is a field of massive vector particle with a intensity of em interaction and the second one is a residue scalar field ( with gradient coupling to charged particles ) with a intensity comparable to that of the strong interaction that is responsible for the symmetry breaking .
let us discuss the above statements in a more concrete fashion using the language of the previous subsection .
first the fock space of the system at the tree level is divided into subspaces , each of which contains particles with the same charge , say @xmath70 , that is spontaneously broken . denoting the total em
charge current density as @xmath73 with @xmath74 the contribution of the charge that is spontaneously broken and @xmath75 the remaining charges that are unbroken . in the symmetry breaking phase
, the corresponding vertex @xmath76 for the charge current density @xmath74 can be decomposed in the same form as eq .
[ charge - sprd ] so that the total vertex for the em charge current density @xmath77 has a form @xmath78 + \overline j ^\mu(p+q , p)+\ldots . \label{jdecomp}\ ] ] here , the vertex for the broken charge density operator @xmath74 is decomposed into a sum of the core part and the spreaded part .
the rest of the vertices corresponding to the unbroken charge current density is denoted by @xmath79 .
then , using the method of the previous subsection , it is not hard to show that the fermion
fermion scattering amplitude @xmath59 in the lowest order approximation in the fine structure constant has the following general form @xmath80 where @xmath81 , @xmath82 denote the matrix elements of the em current operator and @xmath49 and @xmath65 are given by eqs .
[ photon - prop ] and [ pfull - prop ] respectively with @xmath83 where @xmath84 is the broken charge of the charged particle .
[ tfi1 ] indicates that only the broken charge components of the fermions are scattered by an exchange of an authentic massive vector excitation , the scattering between the rest components of the em charge of fermions and between the broken component and the unbroken components of the em charge operator of the fermions are mediated by exchange of massive vector excitation with its longitudinal component @xmath72 given by eq .
[ glmn ] removed .
this effective excitation contains a mixture of massive and massless excitation , which can be read out directly from eq .
[ photon - prop ] .
the massless pole in @xmath49 will not be present explicitly on the right hand side of eq .
[ tfi1 ] in the vacuum case since @xmath85 is a conserved current .
this provides us the freedom to modify the @xmath86 term of @xmath49 .
it is quite natural to write eq .
[ tfi1 ] in an equivalent form , namely , @xmath87 where the propagator @xmath49 in eq .
[ tfi1 ] is replaced by the full propagator @xmath65 . from eq .
[ tfi2 ] , it can be seen that in the fermion
fermion em interaction inside the system , the core part of the em charge participate in the interaction between charged particles exchange a massive vector boson .
since the core part of the broken charge current component of the em charge current is not conserved , the formally similar four terms on the right hand side of eq .
[ tfi2 ] has different physical meaning . in the first term ,
the longitudinal component of @xmath65 contributes since @xmath88 is not conserved by itself .
this term can be interpreted as that both the core component and the spread component of the broken charge participate in the fermion
fermion interaction in this particular channel .
the second and third terms on the right hand side of eq .
[ tfi2 ] have quite different physical properties .
this is because @xmath85 is a conserved quantity the longitudinal component of @xmath65 does not actually participate in the fermion
fermion scattering in the vacuum .
this is one of the reasons why it is difficult to detect the effects of the goldstone boson in the relativistic invariant vacuum state of strong interaction . in case of a nucleon or in a nuclear medium , the contraction of @xmath72 with @xmath85 does not give vanishing result ( see the following discussion ) .
so eq . [ tfi1 ] should be used .
the absence of the longitudinal contribution also means that the effects of the massless goldstone boson is observable .
therefore we do not have a full higgs phenomenon ; the gauge symmetry is therefore considered to be `` spontaneously partial broken '' .
let us consider the lowest order ( in @xmath57 ) inclusive semi - leptonic cross section .
the scattering cross section @xmath89 is related to the imaginary part of the corresponding forward compton scattering amplitude @xmath90 , @xmath91 where `` t '' denotes time ordering and @xmath92 is the photon propagator in the presents of a nucleon @xcite .
the amputation of the external photon line _ outside _ of the nucleon is represented by the factor @xmath93 with @xmath94 the photon wave function renormalization constant .
the discussion of subsection [ sec : one - charge - rev ] for the modification of the photon propagator ( see also ref .
@xcite ) is also applicable for @xmath95 .
but there are two major difference .
first , the vacuum state of the strong interaction is an infinite system in spatial extension whereas a nucleon is a finite size system .
so only when the wave length of the photon is much smaller than the size of a nucleon , the photon propagator would has a similar behavior to the vacuum one @xcite .
otherwise , any pole behavior in the propagators is smeared . for the simplicity of the discussion ,
such a smearing effects are not mentioned explicitly in most part of the discussions except in the small @xmath96 region since we are discussing high energy processes in which the wave length of the ( virtual ) photon is much smaller than the size of a nucleon .
second , the vacuum state of the strong interaction is an isoscalar so that the proper self - energy @xmath97 for the vacuum state , @xmath98 \ket{0}\right |_{1pi}\end{aligned}\ ] ] contains no interference term between the current of broken charge and the current of the unbroken charge due to the isospin symmetry of the vacuum state . for the simplicity of the notation , we shall omit the `` 1pi '' ( one photon irreducible ) subscript in the following .
since the proper self - energy @xmath99 for a photon inside a nucleon is induced by the interaction in an isospin unsymmetric background provided by the nucleon @xcite , the general form of the photon proper self - energy inside a nucleon contains two additional terms corresponding to the interference terms between the current of the broken charge and that of the unbroken charge , namely , @xmath100 \ket{n}.\end{aligned}\ ] ] the term for the broken charge of a nucleon is @xmath101 \ket { n } + \ldots , \label{nbbdecomp}\end{aligned}\ ] ] where @xmath102 acts only on @xmath103 in the above equations .
it can be seen that in order to maintain gauge invariance , the correlator contains not only a single massless goldstone boson pole , but also a double massless pole that is absent in the vacuum case .
the second and third interference terms have isospin odd component that is present in a single nucleon state .
similar to eq .
[ nbbdecomp ] , they can be decomposed as @xmath104 \ket{n } + \ldots , \label{isoodd1 } \\
\bra{n } t \overline
j^\mu(x)b^\nu(0 ) \ket { n } & = & \bra{n } \left [ t\overline j^\mu(x)b^\nu_{core}(0 ) + { \widehat q^\nu\over \widehat q^2 } t\overline j^\mu(x ) b_{sprd}(0 ) \right ] \ket{n } + \ldots . \label{isoodd2}\end{aligned}\ ] ] only a single massless goldstone pole is present for these isospin odd terms .
there is an additional important feature compared to the vacuum case due to the fact a nucleon behaves like a medium for the photon .
the matrix element of the spread current in the momentum space should be of the following form @xmath105 instead of the simple form given by eq .
[ charge - sprd ] . here
@xmath106 is the 4-momentum of the nucleon @xmath107 ( the medium ) , @xmath108 is the 4-momentum of a quark .
the coefficient @xmath109 is determined by the nature of the symmetry breaking inside the nucleon .
@xmath109 is expect to decrease at large @xmath110 as @xmath111 which can be estimated by a conventional finite density computation , with @xmath112 certain mass scale characterizing the concentration of a quark matter inside the nucleon and @xmath113 is the mass of the nucleon .
there is a corresponding change to the propagator @xmath49 for the photon in the nucleon , which is also characterized by @xmath109 .
it should not be discussed quantitatively here .
what is important here is the additional term @xmath114 in the definition of the `` longitudinal '' goldstone boson polarization , namely @xmath115 the medium term @xmath114 , together with the goldstone boson , are responsible for the `` extra '' particle production in a high energy semi - leptonic process involving a nucleon . in most of the expressions in the following , instead of @xmath116 ,
@xmath117 is still used for the goldstone boson contributions for notational simplicity . but care should be taken that the contraction of such a `` @xmath118 '' which is actually @xmath119 , with a conserved current should not be replaced by a zero .
two aspects are important here . the first one is related to the phenomenon of missing charge . in certain channels of reaction , like the semi - leptonic processes , only the core charge of the hadronic system couples to the external probes .
since a finite percent of the broken charge of particle is removed from its core component and added to its spreaded component in the symmetry breaking phase of the system , the system appears to these probes to contain less charge than it would normally be expected @xcite .
the second one is connected to the goldstone boson degree of freedom .
let us imagine that a superconducting nucleon is hit by a high energy virtual photon @xmath120 or @xmath121 boson like the one shown in fig .
[ fig : finals].a , then the quarks ( quasi - particles ) start to radiate gluons , photons and goldstone bosons . because @xmath122 couples to the quarks inside the nucleon with a strength of em interaction and @xmath72 couples to the quarks with a strength of the strong interaction which is two order of magnitude larger that the em one , @xmath123 must also contain an corresponding component that couples to the quarks with a strength of the strong interaction in the superconducting phase .
therefore , the total cross section generated in fig .
[ fig : finals].a is larger then that of a typical em interaction in normal situations .
for the same reason , the total cross section generated in fig .
[ fig : finals].b is also larger then that of a typical em interaction .
these two sets of graphs cancel each other if both of them are emitted by a nucleon which also contains up and down quarks . in this case , the higgs mechanism discussed in subsection [ sec : one - charge - rev ] is realized .
the nucleon
nucleon ( nn ) interaction is normal strong interaction plus the normal em interaction mediated by a massive vector photon . since the strength of the em interaction is much smaller than the strong interaction strength , the total cross section produced by the summation of figs .
[ fig : finals].a and [ fig : finals].b is much smaller than that of other strong interaction processes . in the case of charged lepton and nucleon scattering ,
on the other hand , the charged leptons does not couple to the goldstone boson .
so only fig .
[ fig : finals].a is left .
let us write @xmath124 with @xmath65 the massive photon propagator that couples to the quarks with a strength of em interaction .
one can ignore @xmath65 in this case and write @xmath125 in the vacuum , since @xmath72 is only proportional to @xmath126 , @xmath72 can not mediate force between charged leptons and quarks due to the current conservation on the lepton vertex .
but inside a nucleon , @xmath72 is proportional to @xmath127 .
it contains terms like @xmath128 and @xmath129 due to eq .
[ med - bsprd ] .
these terms do not produce a vanishing result when contracted with the lepton current .
in addition , the contraction of the nucleon 4-momentum @xmath106 with the lepton or quark current generate a factor @xmath110 at large photon energy .
therefore , due to the medium effect , the strong interaction force mediated by the goldstone boson is independent of any mass scale even in high energy reactions .
such a strong interaction force does not , however , exist between nucleons
. therefore , _ there appears to be a new type of strong interaction inside a nuclear system when probed by systems other than another nuclear systems if a nucleon ( or a nucleus ) is superconducting_. as it is shown in the following , this new kind of strong interaction force may be identified with the so called `` hard pomeron '' .
this specific mechanism for the explanation of the violation of the froissart bound has several testable predictions which are explicated in the sequel .
one way of observing the partial breaking of the em gauge symmetry inside a hadronic system is to use the lepton
hadron scattering . since leptons do not couple to the corresponding goldstone boson .
the general scattering amplitudes given by eq .
[ tfi1 ] is simplified @xmath130 where @xmath131 is the electro
weak current of the leptons and @xmath132 .
so only the core part of the electro weak current of the hadrons contributes to the semi - leptonic processes .
the spreaded component of the broken charge of the nucleon is invisible .
this is diagrammatically represented by fig .
[ fig : semilp ] .
the process of the inclusive semi - leptonic dis process is connected to the imaginary part of the forward virtual photon @xmath120 and @xmath133- nucleon compton scattering amplitude @xmath134 which can be decomposed according to eq .
[ nbbdecomp ] . the em interaction in the semi - leptonic @xmath135 scattering conserves parity . the kind of parity conserving decomposition of eq . [ distmn ] is written as a sum of parity even invariant amplitudes @xcite @xmath136 u(ps ) , \label{invampdecomp1}\end{aligned}\ ] ] where @xmath137 is the spinor for a nucleon with momentum @xmath138 and spin @xmath139 and @xmath140 ( @xmath141 ) are invariant amplitudes .
if the em gauge symmetry is spontaneously partial broken , then some of these amplitudes with at least one @xmath118 and/or @xmath142 in front carrying the lorentz indices @xmath112 or @xmath143 develop single and double poles , and @xmath144 respectively for a nucleon in more precise sense .
but we shall not make the distinction between them since the difference can be removed by a proper linear combination of @xmath140 .
so the generality of the discussion will not be affected . ] in @xmath14 .
these parity even invariant amplitudes are @xmath145 , @xmath146 and @xmath147 .
@xmath145 is the coefficient of @xmath86 , so apart from others , it constraints the contributions from the fourth term in eq .
[ nbbdecomp ] which allows it to have a double pole in @xmath14 .
@xmath146 and @xmath147 are coefficient of those terms that are linear in @xmath118 or @xmath142 , they contain contributions from the second and third terms of eq .
[ nbbdecomp ] and the second terms of eqs .
[ isoodd1 ] and [ isoodd2 ] .
therefore , the poles of these three invariant amplitudes can be separated out @xmath148 the invariant amplitudes @xmath140 ( @xmath141 ) are not independent of each other due to the gauge invariance expressed in terms of the ward identity @xmath149 which imposes constraints on them , namely , @xmath150 as discussed above that in the semi - leptonic processes , only the core part of the scattering amplitude are observable . the core part of the invariant amplitude @xmath151 is obtained from it by removing the poles in @xmath14 .
so @xmath152 u(ps ) , \label{invampdecomp2}\end{aligned}\ ] ] which , after considering the constraints eqs .
[ hbz1][hbz3 ] , can be reduced to @xmath153 with @xmath154 where @xmath155 , @xmath156 and @xmath157 depend on @xmath14 slowly .
the total virtual @xmath8 cross section is related to the forward compton scattering amplitude by the optical theorem . for a specific polarization of the incoming ( virtual ) photon or @xmath121 particle with helicity @xmath158 , it is written in the form of @xmath159 where @xmath160 ( @xmath161 ) and @xmath162 as the flux factor , which is not unique when @xmath163 , is chosen to be @xmath164 .
@xmath165 is defined as @xmath166\ket{ps}.\ ] ] it has the same lorentz structure as @xmath167 , namely , @xmath168 optical theorem relates the imaginary part of the invariant compton amplitudes to the structure functions in the dis @xmath169 the laboratory kinematics is defined in the usual way . the plane formed by the 3-momenta of the incident and scattering leptons
is called the scattering plane . the direction defined by @xmath170
is called the 3rd ( or z ) axis ; the direction perpendicular to the the scattering plane is called the 2nd ( or y ) axis ; the direction normal to the plane formed by the 2nd and 3rd axis is called the 1st ( or x ) axis .
the positive quantity @xmath171 is used in the following discussion .
the virtual photon 4-momentum is @xmath172 .
the polarization vector with property @xmath173 and @xmath174 are @xmath175 there is a fourth polarization state of the photon in the spontaneous em gauge symmetry breaking phase .
this state is some times called the longitudinal mode in the literature .
the longitudinal polarization is proportional to @xmath118 , namely , @xmath176 it does not contribute to the semi - leptonic processes interested in this paper .
the unpolarized total virtual @xmath8 cross sections are @xmath177 .
\label{sigmal}\end{aligned}\ ] ] the difference between polarized ones with the photon and nucleon spin polarization antiparallel and parallel respectively is @xmath178 , \label{pol - sig1}\ ] ] and the interference cross section between `` longitudinal '' and transverse photon polarization that is perpendicular to the scattering plane is @xmath179 , \label{pol - sig2}\ ] ] where @xmath180 here @xmath181 the laboratory frame inclusive @xmath182 or @xmath183 unpolarized scattering cross section are related to these @xmath120 total cross section @xmath184 , \label{dsigmaln}\end{aligned}\ ] ] where @xmath185 with k the 4-momentum of the initial lepton , for the unpolarized cross section . here
@xmath186^{-1},\label{eps - param } \\
\gamma & = & { \alpha_{em}\over 2\pi^2 } { e'\over e } { k\over q^2 } { 1\over 1-\epsilon}.\label{gamma - param}\end{aligned}\ ] ] if the nucleon is polarized along @xmath170 @xmath187 if the nucleon is polarized perpendicular to @xmath170 and parallel to the scattering plane @xmath188 the presence of finite @xmath189 , @xmath190 and @xmath191 is not a sufficient condition for the spontaneous partial breaking of the em gauge symmetry .
in fact , the same form of observable quantities in eqs.[tcore1 ] and [ wcore1 ] with additional terms @xmath192 and @xmath193 can be derived from a theory in which a global @xmath13 symmetry that is contained in the em local gauge symmetry is explicitly broken instead of spontaneously .
the difference between them lies in more details related to the existence of extra strong interaction channel in the semi - leptonic processes discussed in subsection [ sec : general - pic ] .
the inclusive cross section in a semi - leptonic process are related to the imaginary part of the forward virtual photon and nucleon compton scattering amplitude which is schematically shown in fig .
[ fig : multipart ] .
let us view this amplitude as the self - energy of the photon when propagate inside a nucleon .
if the virtual photon is emitted by a nucleon , then there is a companion goldstone boson with the proper strength that travel with the photon . due to the higgs mechanism , these two companion excitations effectively screen each other by canceling each others strong interaction components discussed in subsection [ sec : general - pic ] so that besides other strongly interacting particles like gluons and quark antiquark pairs , the subsequent photons and goldstone bosons that are generated by them are also of proper portion that they cancel each other s strong interaction effects . on the other hand ,
if the virtual photon is emitted by a lepton , then there is no companion goldstone boson , the passage of this virtual photon will meet greater resistance because its effective interaction with the environment inside the nucleon is in the strong interaction range .
it also would frustrate the portion of the vacuum state inside the nucleon so that a disproportional number of goldstone are generated , which in turn generate more final hadronic states after they are hadronized .
the hadronization of the goldstone bosons ( or the lack of it , see eq . [ gtstrng ] ) are also selective .
they would not hadronize into hadrons made up of up and down quarks , like the nucleons , @xmath194 mesons , etc .
, since the coupling between the goldstone boson is always screened by the virtual photon inside the system . their coupling to other hadrons made up of quarks that do not participate in the symmetry breaking .
these quarks are strange quarks , charm and bottom quarks , etc . according to the simplest scenario of color superconductivity at low density in refs .
therefore the difference between the total cross section of the @xmath195 scattering and the semi - leptonic @xmath135 scattering should lie in the difference in the contributions of the strange and charm / bottom final state like the @xmath196 meson , @xmath197 mesons , etc . , not in the regular hadronic states made of up and down quarks , like the @xmath194 , @xmath198 , @xmath54 mesons .
the partial spontaneous breaking of the em gauge symmetry inside of a nucleon is characterized by three non - vanishing pole strength of the longitudinal components of the scattering amplitude eqs .
[ h2decomp][h7decomp ] for _ semi - leptonic processes_. since these pole terms are absent for the em interaction part of nucleon nucleon scattering due to the realization of the higgs mechanism in such a system , it would be natural to assume that the effects of the extra particles produced in a semi - leptonic @xmath135 scattering compared to the @xmath195 scattering are encoded in @xmath155 , @xmath156 and @xmath157 , which should increase with the photon energy faster than that are allowed by the froissart bound valid for the @xmath195 interaction . according to the phenomenological success of the regge asymptotics ,
the high energy behavior of @xmath199,@xmath200 , @xmath201 , @xmath202 , @xmath203 , @xmath204 , @xmath205 and @xmath206 follows that of the regge asymptotics with the soft pomeron as the leading trajectory .
it is known that the soft pomeron has an intercept of @xmath207 . for the believers of two pomerons ,
phenomenology @xcite implies that @xmath208 , with @xmath209 the intercept of the `` hard pomeron '' of order 1.4 , for sufficiently large @xmath7 .
let us take the two pomeron hypothesis in the present context and derive its consequences in the following . in the regge theory
, it would be natural to assume that the high energy behavior of the leading singularity term @xmath210 to have the following high energy behavior @xmath211 the high energy behavior of @xmath212 is not known from experimental observations due to the domination of @xmath210 in the cross section . by assuming the universality of the regge asymptotics even in the case of hard pomeron , one can write @xmath213 although there is no strong reason in doing so .
@xmath210 is an isoscalar and @xmath212 contains isoscalar and isovector component . the high energy behavior of @xmath212 can in principle be extracted from the data .
but as a working hypothesis , i shall assume that the isovector component of @xmath212 also satisfy eq .
[ wa - hi - e ] .
the @xmath214 term is related to the amplitudes for @xmath215 with the two transversely polarized photon having opposite helicity in the center of mass frame .
the isoscalar component of each of them have the following asymptotic behavior at hight energies in the @xmath216 channel , namely , @xmath217 where @xmath218 is the intercept of the soft pomeron and the leading trajectory for @xmath219 is assumed to be that of the hard pomeron also .
there is no general argumentation which allows us to write down the high energy behavior for their isovector components . in principle , the energy dependence of @xmath189 , @xmath190 and @xmath191 and their imaginary parts may have a completely different @xmath143 dependence . for fixed @xmath7
, assuming the asymptotic behavior eqs .
[ wb - hi - e ] , [ wa - hi - e ] and [ zcs ] hold , crossing symmetry requires that the corresponding amplitudes are of the following general form @xmath220 where @xmath221 are polynomials of @xmath143 .
the @xmath7 dependence in the above equations are suppressed .
@xmath222 and possibly non - vanishing is known ( see the following discussion ) from the study of the modification of gdh sum rule @xcite motivated by analysis of experimental data .
the form of @xmath223 remains to be determined .
the isospin structure of @xmath189 , @xmath190 and @xmath191 and their imaginary parts can be found under the scenario of refs . @xcite and @xcite .
parameter @xmath190 is the strength of the double pole in @xmath14 in the compton scattering amplitudes ; it corresponds to the last term of eq .
[ nbbdecomp ] .
since the broken charge is an isoscalar , @xmath190 is also an isoscalar .
parameter @xmath189 and @xmath191 are the strength of the single pole in @xmath14 in the compton scattering amplitude , they contain contributions from the second and third terms in eq .
[ nbbdecomp ] and the second term in eqs .
[ isoodd1 ] and [ isoodd2 ] .
therefore @xmath224 so , the most singular piece of the compton amplitude in @xmath14 can be observed in its isoscalar component and the next singular pieces of the compton amplitude manifested themselves in the isovector component of its amplitude .
let us assume that all of the structure functions considered are regular in @xmath7 when @xmath7 is small , then following leading @xmath226 behavior at small @xmath7 hold @xmath227 , \label{wlasmp}\end{aligned}\ ] ] which incorporate the possibility of spontaneous partial breaking of the em gauge symmetry by assuming a non - vanishing @xmath228 . here
all terms of order @xmath7 or higher are dropped except those ones that depend on @xmath212 and @xmath210 since they have a quite different energy dependence from the normal ones . if @xmath229 and @xmath230 , then the following limiting behavior follows @xmath231,\label{sigmalasymp } \\
\sigma_{lt } & \to & { 4\pi^2\alpha_{em}\over m^2 } { \sqrt{q^2}\over k } \left [ -{\nu^2\over q^2 } z_c \right ] , \label{sigmaltasymp}\end{aligned}\ ] ] when @xmath7 is small .
the longitudinal total virtual photon
nucleon scattering cross sections @xmath232 and @xmath233 behave like @xmath226 according to eqs .
[ sigmalasymp ] and [ sigmaltasymp ] .
such a behavior can not continue all the way down to @xmath234 since a nucleon is a finite system , such a singular behavior can not be fully developed in a finite system .
the rise of the cross sections as @xmath225 is saturated at certain small @xmath235 value with @xmath236 decreasing function of the photon energy @xmath143 @xcite .
the simplest way to represents the finite size ( of the nucleon ) effects is to replace @xmath226 in the pole terms by the following ad hoc form @xmath237 for the definiteness of the discussion .
it turns to @xmath238 instead of infinity in the @xmath225 limit .
in addition , according to the discussions of ref .
@xcite , the maximum of the peak at the @xmath234 position is proportional to the maximum number of the coherent interaction of the photon with the finite medium at large @xmath143 .
so we can further choose a specific form for @xmath236 @xmath239 as a working hypothesis with parameters @xmath47 and @xmath240 slow varying function of @xmath143 .
it must be emphasized that the functional forms given in eqs .
[ q2smear ] and [ q0 ] are chosen only based on qualitative considerations .
other forms having the same qualitative properties may be better when detailed fitting to the experimental data are made . therefore , in the @xmath225 limit
, one should replace @xmath189 , @xmath190 and @xmath191 by @xmath241 and their imaginary part by @xmath242 after the substitutions eqs .
[ a - repl][wc - repl ] are made in eqs .
[ dsigmaln ] , [ dsigmaln - p1 ] and [ dsigmaln - p2 ] , it is easily seen that terms involving @xmath228 and @xmath219 are actually higher order terms in @xmath7 than the leading term .
the unpolarized @xmath8 cross section is @xmath243 with @xmath244 therefore there seems to be no anomalies here .
the real photon
nucleon total cross section has the same form as the one in which the em gauge symmetry is not broken .
the high energy behavior of @xmath202 obtained from @xmath146 by subtracting the possible pole term in @xmath7 is expected to follow the regge asymptotics : @xmath245 with the leading trajectory that of the soft pomeron having an intercept @xmath246 . by assuming that the imaginary part of @xmath202 has the same large energy asymptotics , one get @xmath247 it agrees with the experimental observation @xcite well . the polarized @xmath248 cross section
are @xmath249 the first of the above equations is studied in the context of the apparent violation of the gdh sum rule @xcite , which states @xmath250 with @xmath251 the nucleon anomalous magnetic moment and @xmath252 the inelastic threshold in @xmath143 . if the em gauge symmetry is spontaneously partial broken , then a modification of this sum rule @xcite is @xmath253 in the notation of this paper . the parameter @xmath254 defined in @xcite
is related to the polynomial @xmath255 in eq .
[ c - disp ] in the following way @xmath256 where @xmath257 .
so finite modification of the gdh sum rule constrained by observation restricts the polynomial @xmath255 to be at most a finite constant @xmath258 . 1 .
the current analysis of published data @xcite indicates that @xmath259 and mostly isovector .
the most recent experimental study of gdh sum rule at mami in mainz , germany reduced the discrepancy between the theory and data to within @xmath260 for a proton based on a preliminary analysis of the data taken .
similar study on a neutron , which is relevant to the isovector component of the difference that can even be a violation in sign , is still lacking .
the theoretical extrapolation of the dis data @xcite , which includes all possible photon energies also has a less than @xmath260 difference for the gdh sum rule . therefore whether or not the gdh sum rule is in fact violated is still an open question .
2 . since according to eq .
[ eq3 ] , c contains both the isoscalar and isovector components , the fact that the isoscalar component gets suppressed can also be qualitatively understood .
this is because the isoscalar component relates to the second and third terms in eq .
[ nbbdecomp ] , which contain @xmath261 that are reduced in strength compared to its value in the normal phase in which em gauge symmetry is kept . on the other hand
, the isovector component of @xmath191 relates to the second terms in eqs .
[ isoodd1 ] and [ isoodd2 ] , which do not contain @xmath261 and therefore is not suppressed .
3 . before continuing , a discussion regarding the subtle nature of @xmath219 term in eq .
[ sigp0 ] is necessary . due to the separation of the pole terms from the invariant functions @xmath140 ( @xmath141 ) in the forward compton scattering amplitudes , the intermediate results in deriving the modified gdh sum rule is slightly different from those in ref .
@xcite . here
@xmath262 corresponds to @xmath263 of ref .
the dominant contribution to the left hand side of the gdh sum rule is from the photo pion production through the @xmath264 resonance .
@xmath219 does not contain such contributions however .
the reason is the pion and any other mesons made of up and down quarks belong to the hadronic subspace in which the higgs mechanism fully operates . since @xmath219 is the contribution of the goldstone boson , it is only significant beyond the @xmath196 production threshold .
the @xmath196 and @xmath197 contains no up or down quarks so that they do not couple to the goldstone boson . since the goldstone boson couples to the hadronic matter with a strength of strong interaction
, the lack of the contribution of the goldstone boson means that the em production of @xmath196 and @xmath197 contain an uncanceled extra component with the rate comparable to that of the strong interaction if the em gauge symmetry is spontaneously partial broken .
therefore the effect of @xmath219 on the left hand side of gdh sum rule can be ignored below the @xmath196 production threshold , which is greater than 1 gev , because it is of higher order in @xmath57 .
the contribution of these states to the sum rule are expected to be small due to the high threshold and the suppression imposed by okubo
iuzuka ( ozi ) rule @xcite .
the small @xmath7 region is defined as the region in which the pole behaviors of @xmath189 , @xmath190 and @xmath191 is not dominant , namely , the region in which @xmath7 is not much larger than @xmath236 .
this is a transition region in which the effects of the symmetry breaking terms @xmath189 , @xmath190 and @xmath191 and all other normal terms compete with each other due to the finite size effects of a nucleon .
the more detailed description of this region is beyond the scope of this work , which is based mainly on the symmetry considerations .
this is a region that can be experimentally studied in a more quantitative way .
in the bjorken limit defined as the kinematic region in which @xmath265 , @xmath266 and @xmath267fixed , the quark substructure of a nucleon begin to reveal itself in the structure functions of the nucleon by exhibiting scaling . this is because the large @xmath7 virtual photon @xmath120 interacts with a quark inside a nucleon with a ( light cone ) time duration much smaller than the soft and coherent processes that symmetry breaking and confinement of quarks take place .
thus the virtual photon provides a snapshot of the quark distribution of a nucleon on the light cone .
combined with the asymptotic freedom , this leads to the parton model in which a factorization of the leading twist hard processes and the much slower soft processes can be factorized . in the region where @xmath7 is much smaller than @xmath143 , which means small @xmath0 , the light cone time duration in the current
current correlator becomes much larger than the virtual photon s wave length , the coherent processes can accumulate their strength @xcite .
therefore one expects 1 ) the manifestation of the leading twist coherent subprocesses due to symmetry breaking , if any , in these high energy reactions and 2 ) non - leading twist contributions .
the scaling properties of the structure functions in the bjorken limit make it useful to write them as @xmath268,\label{g1}\\ g_2(x , q^2 ) & = & { \nu^2\over m^2 } z_2(-q^2,\nu).\label{g2}\end{aligned}\ ] ] true scaling means that these structure functions are independent of @xmath7 at large enough @xmath0 . in qcd , they change only slowly as a function of @xmath7 .
such a scaling behavior of the above defined structure functions has a rather natural interpretation in terms of parton model .
in addition to these conventional structure functions , a set of new ones characterizing the symmetry breaking are defined in the following @xmath269 since , as it is shown above , the physical processes interested here due to the spontaneous partial breaking of the em gauge symmetry _ does not depend on a particular finite mass scale _ , it is expected that scaling behavior should be present in certain forms at high energies .
_ it is assumed that these new structure functions change slowly with @xmath7 in a logarithmic fashion in the paper for the further development of the ideas that can be experimentally examined_. under such an assumption , it is expected that they are only significant in the small @xmath0 kinematic region as functions of @xmath0 in which coherent processes starts to build up .
the unpolarized differential cross section for the inclusive processes is @xmath270 f_2(x , q^2 ) -
y^2 f_l(x , q^2 ) \right \}. \label{disunpol}\ ] ] the polarized differential cross section for the inclusive process proportional to the polarized total virtual photon and nucleon cross sections in eqs .
[ dsigmaln - p1 ] and [ dsigmaln - p2 ] .
@xmath271 and @xmath233 are related to the spin structure functions @xmath6 and @xmath272 in the following way @xmath273 \label{sltg1g2}\end{aligned}\ ] ] with @xmath274 due to the presence of the quite singular term @xmath275 in eq .
[ sltg1g2 ] , if symmetry breaking allows for @xmath276 , then the structure functions @xmath6 and @xmath272 at small @xmath0 can not be extracted from the observables @xmath277 instead , only the following combination can be extracted @xmath278 in the bjorken limit , we have @xmath279 namely , it is impossible to extract @xmath272 at small @xmath0 from the data on @xmath280 and @xmath281 even in the true bjorken limit .
the diffractive meson production in the dis processes provide additional means for the check various theoretical pictures for the dis at small @xmath0 .
the pomeron exchange picture is proven to be a good one to describe the data @xcite .
the testable signature for the scenario proposed in this work are the following 1 . according to the scenario proposed in this work ,
the diffractive production of the @xmath194 meson , which is made up of up and down quarks , is dominated by the soft pomeron with an intercept of @xmath282 .
this is because , as discussed above , the hilbert space spanned by the goldstone boson does not contribute to the @xmath194 meson production .
the semi - leptonic electro- diffractive production of the @xmath196 and @xmath197 mesons , on the other hand , is dominated by the hard pomeron with an intercept of @xmath283 since the strange and charm quarks , like the charged leptons , have zero nucleon charge .
so the cross section for the process is described by the imaginary part of fig .
[ fig : multipart].b , which , due to the ( lack of ) the contribution of the goldstone boson in the up and down quark subspace , couples to the nucleon with the strength of strong interaction . the change of the diffractive cross section for the @xmath196 and @xmath197 production should follow that of the hard pomeron if there is a spontaneous partial breaking of the em gauge symmetry according to the scenario proposed here .
3 . as indicated by eq .
[ jsprd - jcore ] , the core part of the em current operator for the up and down quarks is not conserved in the symmetry breaking case , it follows from eq .
[ semi - lp - t ] that current conservation shall _ appear _ to be violated in exclusive processes like the diffractive meson production . 1 .
if the rapid increase of measured @xmath5 is attributed to the symmetry breaking terms @xmath212 and @xmath210 , which are allowed to increase beyond the froissart bound like eq . [ wb - hi - e ] and [ wa - hi - e ] , then we have observed behavior @xmath284 at small @xmath0 .
it agrees with the hard pomeron interpretation given here .
the structure function @xmath285 is extracted from the measured unpolarized cross section corresponding to the left hand side of eq .
[ disunpol ] . in order to get @xmath5
, longitudinal structure function @xmath286 at small @xmath0 , which is believed to be small , has to be known .
the next - to - leading order perturbative qcd calculation of @xmath286 is shown to incorporate the data , but it could be a self - consistent game @xcite in the small @xmath0 region .
there could be other values for @xmath5 and @xmath286 that also describe the data on the left hand side of eq .
[ disunpol ] .
independent experimental determination of @xmath286 is desirable . for fixed initial lepton energy experiment , in which the quantity @xmath287 is related to @xmath7 by @xmath288 with @xmath289 and @xmath290 the energy of the incident charged lepton ,
it is difficult to measure @xmath286 since the @xmath7 dependences of both @xmath5 and @xmath286 are also a measured quantities .
3 . if the experiment can be designed to take data with fixed @xmath7 and @xmath0 , then the @xmath287 dependence of the cross section on the left hand side of eq .
[ disunpol ] , which is only quadratic in @xmath287 , can be used to determine both @xmath5 and @xmath286 independently .
of special interest to the scenario proposed in this work is the small @xmath0 region where @xmath5 is shown to rise quickly . if it is indeed that the symmetry breaking terms @xmath212 and @xmath210 are causing the quick changes of @xmath5 at small @xmath0 , then @xmath5 and @xmath286 at small enough @xmath0 will be @xmath291 let us further suppose that @xmath292 at small enough @xmath0 , then the right hand side of eq .
[ disunpol ] will be a linear rather than quadratic function of @xmath287 .
the violation of the gottfried sum rule is related to the unexpected behavior of the isospin odd component of @xmath5 .
it is believed to be caused by the small @xmath0 region of @xmath285 . according to eq .
[ f2behv ] , the symmetry breaking effects , if present , do not contribute to it since @xmath210 is an isospin even term ( see eq . [ eq2 ] ) .
but this idealized situation was not realized in the analysis of the data because @xmath286 is in fact unknown from direct observation .
the extracted @xmath293 under certain theoretical assumption may contain small contribution of the true @xmath286 , namely @xmath294 with @xmath295 a small number that can has a @xmath0 dependence . since @xmath286 contains @xmath212 , which contains an isospin odd component ( see eq .
[ eq1 ] ) , the extracted @xmath296 contains a term that violate the gottfried sum rule . but
this violation is expected to diminish when more information about @xmath286 at small @xmath0 is known according to the scenario proposed here . since from eq .
[ wa - hi - e ] , @xmath297 , the isospin odd component of @xmath296 contains a small piece that decreases to zero very slowly compared to the small @xmath0 extrapolation used to find the discrepancy
. this could be the source of the violation of the gottfried sum rule .
+ .the effective charges for the electro - weak couplings in the dis between leptons and a nucleon / nucleus according to the standard model .
@xmath298 and @xmath299 are coefficients of the vector and axial vector current operators in the hadronic weak neutral current operator . here
@xmath300 is the weinberg angle .
`` @xmath301 '' and `` @xmath302 '' denote up and down quarks respectively .
the corresponding charge for an anti - quark is just opposite .
the value for @xmath303 in the table depends on the color and the momentum fraction @xmath0 of the corresponding quark .
[ tab : compare ] [ cols="^,^,^,^,^",options="header " , ] 5 .
the difference between the charge lepton dis data and the neutrino dis data at small @xmath0 ( @xmath304 ) remains after heavy target corrections are included @xcite .
such a difference is anticipated @xcite .
the reason is because the isoscalar component of the em charge ( the nucleon number density current ) of the up and down quarks contributes differently to the charge lepton neutral current dis , neutrino neutral current and neutrino charged current dis . for the charged lepton neutral current dis , the isoscalar charge is @xmath305 .
it is @xmath306 for neutrino neutral current dis , which is smaller in magnitude ( see table [ tab : compare ] ) . here , @xmath300 is the weinberg angle .
the isoscalar charge does not contribute to the neutrino charged current dis processes .
+ the structure function @xmath5 is extracted from the experimental data based on the assumption that the core charge of quark is not modified .
so the reduction of the core charge of the light quarks in the superconducting phase is expected to reduce the extracted structure function against its true value .
such a reduction is larger for the results from charge lepton neutral current dis than the one from neutrino neutral current dis .
there is no reduction of the extracted structure function in the neutrino charged current dis .
the results for @xmath5 extracted from the ccfr neutrino dis experiment and the ones obtained from various charged lepton neutral current dis experiment do indeed show such a tendency .
the violation of the charge symmetry in the unpolarized structure function @xmath5 is due to non - linear effects .
the violation is derived based on the assumption that the inclusive cross section at small @xmath0 can be written as @xmath307 where the summation is over all possible flavors and @xmath308 is _ independent of the charge @xmath309_. + this assumption is true in the normal cases since the fine structure constant @xmath57 for em interaction is less than @xmath310 ; higher order em correction to @xmath5 can be neglected . + it is not true in case that the em gauge symmetry is partially broken for a nucleon .
take fig.[fig : multipart].b for example , the em induced high order _ effective _ em interaction between the upper and lower blocks can not be ignored .
this is because the propagator @xmath49 contains the strong interaction component @xmath311 ( see eq .
[ gtstrng ] ) . according to ref .
@xcite , @xmath312 with @xmath84 the broken charge .
but eqs . [ tfi1 ] and [ semi - lp - t ] tell us that the coupling mediated by @xmath49 is proportional to the product of the core part of the whole em charges of the two interacting quarks inside the nucleon .
so the strength of the force mediated by @xmath49 in this case is in the strong interaction range .
higher order effects are important .
the magic of the partial breaking of the em gauge symmetry inside a nucleon is to make @xmath308 at small @xmath0 to depend on the relative charge @xmath313 due to the unscreened contribution of the goldstone boson , namely @xmath314 with @xmath315 and @xmath316 the core part of the quark s em charge .
+ if rigorous charge symmetry condition @xmath317 is assumed , where the `` @xmath318 '' denotes a proton and `` @xmath319 '' denotes a neutron , then the compared quantities in the literature , which are @xmath320 are not zero if the em gauge symmetry is partially broken inside an nucleon .
this is due to the fact that the compared quantities have different arguments .
+ so the finding of refs .
@xcite and @xcite can be interpreted as an indication of the partial breaking of the em gauge symmetry inside a nucleon provided that eq .
[ csym ] is true .
7 . the polarized structure functions @xmath6 and @xmath272 can not be extracted from the dis data even in principle if the symmetry breaking term @xmath276 since it is @xmath321 and @xmath322 ( see eqs .
[ tildeg1 ] and [ tildeg2 ] ) that are related to the experimental observables @xmath280 and @xmath281 . the extracted values for @xmath6 and @xmath272 also contain our uncertainty with the value of @xmath323 at small @xmath0 even both @xmath280 and @xmath281 can be measured separately .
the experimentally extracted @xmath321 contains a small singular piece @xmath324 if the symmetry breaking effects are present .
it is possible that the rapid decrease of @xmath6 for a neutron reported in ref . @xcite and the rapid increase of the same quantity for a proton implied in the original data of ref .
@xcite are manifestations of a finite @xmath275 for both a proton and a neutron . the most recent measurement of @xmath280 at smc for @xmath0 as low as @xmath325 for a proton is plotted in fig .
[ fig : smc ] without the errors been displayed .
the solid line that fits the data has a power law behavior of @xmath326 although the power @xmath327 may not be taken seriously at the present stage of the experimental accuracy , one message is quite certain from that data : @xmath280 does not show a trend that approach to zero when @xmath328 . given @xmath329 as a working hypothesis
, it leads to @xmath330 , which is quite singular . + ( 1200,720)(0,0 )
= cmr10 at 10pt ( 370.0,217.0 ) + ' '' '' + ( 350,217)(0,0)[r]0 ( 1450.0,217.0 ) + ' '' '' + ( 370.0,334.0 ) + ' '' '' + ( 350,334)(0,0)[r]40 ( 1450.0,334.0 ) + ' '' '' + ( 370.0,452.0 ) + ' '' '' + ( 350,452)(0,0)[r]80 ( 1450.0,452.0 ) + ' '' '' + ( 370.0,569.0 ) + ' '' '' + ( 350,569)(0,0)[r]120 ( 1450.0,569.0 ) + ' '' '' + ( 370.0,687.0 ) + ' '' '' + ( 350,687)(0,0)[r]160 ( 1450.0,687.0 ) + ' '' '' + ( 370.0,158.0 ) + ' '' '' + ( 370,117)(0,0)0.0001 ( 370.0,696.0 ) + ' '' '' + ( 474.0,158.0 ) + ' '' '' + ( 474.0,706.0 ) + ' '' '' + ( 535.0,158.0 ) + ' '' '' + ( 535.0,706.0 ) + ' '' '' + ( 579.0,158.0 ) + ' '' '' + ( 579.0,706.0 ) + ' '' '' + ( 612.0,158.0 ) + ' '' '' + ( 612.0,706.0 ) + ' '' '' + ( 640.0,158.0 ) + ' '' '' + ( 640.0,706.0 ) + ' '' '' + ( 663.0,158.0 ) + ' '' '' + ( 663.0,706.0 ) + ' '' '' + ( 683.0,158.0 ) + ' '' '' + ( 683.0,706.0 ) + ' '' '' + ( 700.0,158.0 ) + ' '' '' + ( 700.0,706.0 ) + ' '' '' + ( 716.0,158.0 ) + ' '' '' + ( 716,117)(0,0)0.001 ( 716.0,696.0 ) + ' '' '' + ( 821.0,158.0 ) + ' '' '' + ( 821.0,706.0 ) + ' '' '' + ( 882.0,158.0 ) + ' '' '' + ( 882.0,706.0 ) + ' '' '' + ( 925.0,158.0 ) + ' '' '' + ( 925.0,706.0 ) + ' '' '' + ( 958.0,158.0 ) + ' '' '' + ( 958.0,706.0 ) + ' '' '' + ( 986.0,158.0 ) + ' '' '' + ( 986.0,706.0 ) + ' '' '' + ( 1009.0,158.0 ) + ' '' '' + ( 1009.0,706.0 ) + ' '' '' + ( 1029.0,158.0 ) + ' '' '' + ( 1029.0,706.0 ) + ' '' '' + ( 1047.0,158.0 ) + ' '' '' + ( 1047.0,706.0 ) + ' '' '' + ( 1063.0,158.0 ) + ' '' '' + ( 1063,117)(0,0)0.01 ( 1063.0,696.0 ) + ' '' '' + ( 1167.0,158.0 ) + ' '' '' + ( 1167.0,706.0 ) + ' '' '' + ( 1228.0,158.0 ) + ' '' '' + ( 1228.0,706.0 ) + ' '' '' + ( 1271.0,158.0 ) + ' '' '' + ( 1271.0,706.0 ) + ' '' '' + ( 1305.0,158.0 ) + ' '' '' + ( 1305.0,706.0 ) + ' '' '' + ( 1332.0,158.0 ) + ' '' '' + ( 1332.0,706.0 ) + ' '' '' + ( 1355.0,158.0 ) + ' '' '' + ( 1355.0,706.0 ) + ' '' '' + ( 1375.0,158.0 ) + ' '' '' + ( 1375.0,706.0 ) + ' '' '' + ( 1393.0,158.0 ) + ' '' '' + ( 1393.0,706.0 ) + ' '' '' + ( 1409.0,158.0 ) + ' '' '' + ( 1409,117)(0,0)0.1 ( 1409.0,696.0 ) + ' '' '' + ( 370.0,158.0 ) + ' '' '' + ( 1470.0,158.0 ) + ' '' '' + ( 370.0,716.0 ) + ' '' '' + ( 920,76)(0,0)@xmath0 ( 161,437)(0,0)[l]@xmath331 ( 920,521)(0,0)[l]@xmath332 ( 370.0,158.0 ) + ' '' '' + ( 370,687 ) ( 489,417 ) ( 575,232 ) ( 647,198 ) ( 716,278 ) ( 787,299 ) ( 854,238 ) ( 936,212 ) ( 1025,227 ) ( 1116,235 ) ( 1197,225 ) ( 1249,226 ) ( 1301,218 ) ( 1370,226 ) ( 1438,225 ) ( 370,686 ) ( 370.58,679.40)(0.492,-1.911)19 + ' '' '' + ( 369.17,682.70)(11.000,-37.698)2 + ' '' '' + ( 381.58,639.00)(0.492,-1.722)19 + ' '' '' + ( 380.17,642.00)(11.000,-34.000)2 + ' '' '' + ( 392.58,602.45)(0.492,-1.581)19 + ' '' '' + ( 391.17,605.23)(11.000,-31.226)2 + ' '' '' + ( 403.58,568.91)(0.492,-1.439)19 + ' '' '' + ( 402.17,571.45)(11.000,-28.453)2 + ' '' '' + ( 414.58,538.71)(0.492,-1.186)21 + ' '' '' + ( 413.17,540.86)(12.000,-25.855)2 + ' '' '' + ( 426.58,510.66)(0.492,-1.203)19 + ' '' '' + ( 425.17,512.83)(11.000,-23.830)2 + ' '' '' + ( 437.58,484.96)(0.492,-1.109)19 + ' '' '' + ( 436.17,486.98)(11.000,-21.981)2 + ' '' '' + ( 448.58,461.41)(0.492,-0.967)19 + ' '' '' + ( 447.17,463.21)(11.000,-19.207)2 + ' '' '' + ( 459.58,440.57)(0.492,-0.920)19 + ' '' '' + ( 458.17,442.28)(11.000,-18.283)2 + ' '' '' + ( 470.58,420.87)(0.492,-0.826)19 + ' '' '' + ( 469.17,422.43)(11.000,-16.434)2 + ' '' '' + ( 481.58,403.17)(0.492,-0.732)19 + ' '' '' + ( 480.17,404.58)(11.000,-14.585)2 + ' '' '' + ( 492.58,387.32)(0.492,-0.684)19 + ' '' '' + ( 491.17,388.66)(11.000,-13.660)2 + ' '' '' + ( 503.58,372.47)(0.492,-0.637)19 + ' '' '' + ( 502.17,373.74)(11.000,-12.736)2 + ' '' '' + ( 514.58,358.79)(0.492,-0.539)21 + ' '' '' + ( 513.17,359.89)(12.000,-11.893)2 + ' '' '' + ( 526.00,346.92)(0.496,-0.492)19 + ' '' '' + ( 526.00,347.17)(9.962,-11.000)2 + ' '' '' + ( 537.00,335.92)(0.496,-0.492)19 + ' '' '' + ( 537.00,336.17)(9.962,-11.000)2 + ' '' '' + ( 548.00,324.93)(0.611,-0.489)15 + ' '' '' + ( 548.00,325.17)(9.778,-9.000)2 + ' '' '' + ( 559.00,315.93)(0.611,-0.489)15 + ' '' '' + ( 559.00,316.17)(9.778,-9.000)2 + ' '' '' + ( 570.00,306.93)(0.692,-0.488)13 + ' '' '' + ( 570.00,307.17)(9.651,-8.000)2 + ' '' '' + ( 581.00,298.93)(0.798,-0.485)11 + ' '' '' + ( 581.00,299.17)(9.488,-7.000)2 + ' '' '' + ( 592.00,291.93)(0.798,-0.485)11 + ' '' '' + ( 592.00,292.17)(9.488,-7.000)2 + ' '' '' + ( 603.00,284.93)(0.943,-0.482)9 + ' '' '' + ( 603.00,285.17)(9.270,-6.000)2 + ' '' '' + ( 614.00,278.93)(1.267,-0.477)7 + ' '' '' + ( 614.00,279.17)(9.800,-5.000)2 + ' '' '' + ( 626.00,273.93)(1.155,-0.477)7 + ' '' '' + ( 626.00,274.17)(8.966,-5.000)2 + ' '' '' + ( 637.00,268.93)(1.155,-0.477)7 + ' '' '' + ( 637.00,269.17)(8.966,-5.000)2 + ' '' '' + ( 648.00,263.94)(1.505,-0.468)5 + ' '' '' + ( 648.00,264.17)(8.509,-4.000)2 + ' '' '' + ( 659.00,259.94)(1.505,-0.468)5 + ' '' '' + ( 659.00,260.17)(8.509,-4.000)2 + ' '' '' + ( 670.00,255.95)(2.248,-0.447)3 + ' '' '' + ( 670.00,256.17)(7.748,-3.000)2 + ' '' '' + ( 681.00,252.94)(1.505,-0.468)5 + ' '' '' + ( 681.00,253.17)(8.509,-4.000)2 + ' '' '' + ( 692,248.17 ) + ' '' '' + ( 692.00,249.17)(6.226,-2.000)2 + ' '' '' + ( 703.00,246.95)(2.248,-0.447)3 + ' '' '' + ( 703.00,247.17)(7.748,-3.000)2 + ' '' '' + ( 714.00,243.95)(2.472,-0.447)3 + ' '' '' + ( 714.00,244.17)(8.472,-3.000)2 + ' '' '' + ( 726,240.17 ) + ' '' '' + ( 726.00,241.17)(6.226,-2.000)2 + ' '' '' + ( 737,238.17 ) + ' '' '' + ( 737.00,239.17)(6.226,-2.000)2 + ' '' '' + ( 748,236.17 ) + ' '' '' + ( 748.00,237.17)(6.226,-2.000)2 + ' '' '' + ( 759,234.67 ) + ' '' '' + ( 759.00,235.17)(5.500,-1.000)2 + ' '' '' + ( 770,233.17 ) + ' '' '' + ( 770.00,234.17)(6.226,-2.000)2 + ' '' '' + ( 781,231.67 ) + ' '' '' + ( 781.00,232.17)(5.500,-1.000)2 + ' '' '' + ( 792,230.17 ) + ' '' '' + ( 792.00,231.17)(6.226,-2.000)2 + ' '' '' + ( 803,228.67 ) + ' '' '' + ( 803.00,229.17)(5.500,-1.000)2 + ' '' '' + ( 814,227.67 ) + ' '' '' + ( 814.00,228.17)(6.000,-1.000)2 + ' '' '' + ( 826,226.67 ) + ' '' '' + ( 826.00,227.17)(5.500,-1.000)2 + ' '' '' + ( 837,225.67 ) + ' '' '' + ( 837.00,226.17)(5.500,-1.000)2 + ' '' '' + ( 848,224.67 ) + ' '' '' + ( 848.00,225.17)(5.500,-1.000)2 + ' '' '' + ( 870,223.67 ) + ' '' '' + ( 870.00,224.17)(5.500,-1.000)2 + ' '' '' + ( 881,222.67 ) + ' '' ''
+ ( 881.00,223.17)(5.500,-1.000)2 + ' '' '' + ( 859.0,225.0 ) + ' '' '' + ( 903,221.67 ) + ' '' '' + ( 903.00,222.17)(5.500,-1.000)2 + ' '' '' + ( 892.0,223.0 ) + ' '' '' + ( 926,220.67 ) + ' '' '' + ( 926.00,221.17)(5.500,-1.000)2 + ' '' '' + ( 914.0,222.0 ) + ' '' '' + ( 959,219.67 ) + ' '' '' + ( 959.00,220.17)(5.500,-1.000)2 + ' '' '' + ( 937.0,221.0 ) + ' '' '' + ( 992,218.67 ) + ' '' '' + ( 992.00,219.17)(5.500,-1.000)2 + ' '' '' + ( 970.0,220.0 ) + ' '' '' + ( 1048,217.67 ) + ' '' '' + ( 1048.00,218.17)(5.500,-1.000)2 + ' '' '' + ( 1003.0,219.0 ) + ' '' '' + ( 1148,216.67 ) + ' '' '' + ( 1148.00,217.17)(5.500,-1.000)2 + ' '' '' + ( 1059.0,218.0 ) + ' '' '' + ( 1159.0,217.0 ) + ' '' '' + however such a singular behavior for @xmath6 is of high twist effects , it is expected to go a way in the @xmath333 limit .
the contribution of @xmath275 to @xmath334 is not suppressed in the bjorken limit .
in addition it is more singular than @xmath335 since @xmath336 .
the spin `` crisis '' is still with us .
it can mean either 1 ) the sea quark of a nucleon is negatively polarized against the nucleon spin or / and 2 ) the orbital angular momentum of the quarks is finite or / and 3 ) the contribution of the gluons polarization through the anomalous term or / and 4 ) the strange content of a nucleon is finite to provide the rest of the first moment @xmath337 .
the recent measurement of the sea quark polarization turns out to be quite small @xcite .
however there is still a lack of the gluons polarization @xmath338 data at the present .
it is possible that such a picture is correct for @xmath6 .
but the recently found rapid change of @xmath339 at small @xmath0 could make the crisis even more severe .
+ the assumption of a partial spontaneous breaking of the em gauge symmetry could provide at least a partial understanding for the spin crisis related problems .
this is because the extracted @xmath321 contains a singular piece @xmath340 , which should give a divergent result for the first moment @xmath337 as the lower bound of the @xmath0 integration is getting smaller and smaller in future measurements .
the reason the present value of @xmath337 to be finite is due to the assumed normal regge behavior extrapolation from the known data at larger @xmath0 .
the bjorken sum rule should be respected at least in the bjorken limit .
it is not clear whether or not the bjorken sum rule is violated by the experimental data at finite @xmath143 and @xmath7 due to the presence of @xmath275 in the extracted @xmath321 in the case of symmetry breaking when more data for @xmath321 at smaller @xmath0 is known .
the reason that the bjorken sum rule appears to be not violated in the current experimental data could mean either that it is in fact not violated or it is not violated if a normal regge behavior for @xmath6 at small @xmath0 is assumed . in either case , it only mean that the bjorken sum rule is respected by the regular @xmath6 .
+ it is highly possible the the bjorken sum rule is violated by @xmath341 if more experimental information about the its small @xmath0 behavior is known . in that case , the isospin odd component of the imaginary part of @xmath191 or @xmath219 is different from zero , which is allowed under the scenario proposed here ( see eq .
[ eq3 ] ) .
but it must be emphasized that the possible violate of the isospin odd component of the gdh sum rule and the possible violation of the bjorken sum rule is not in one to one correspondence even in symmetry breaking case .
the violation of the gdh sum rule is due to the asymptotic polynomial part of @xmath191 defined in eq .
[ c - disp ] , which does not contribute to quantities involved in the dis cross sections .
it may or may not be zero even in the symmetry breaking case . 10 .
for the same reason as the ones given above , the burkhardt
cottingham sum rule is likely to be violated when more data about @xmath281 ( or @xmath334 ) is known at small @xmath0 due to the singular piece proportional to @xmath275 .
it will remain to be so even in the true bjorken limit ( see eq .
[ g2-blmt ] ) if the nucleon is superconducting .
the energy dependence of the @xmath194 electroproduction follows that of the soft pomeron dominance in the observation .
this is expected even in the symmetry breaking case .
the rapid increase of the @xmath196 and @xmath197 production cross section @xcite that is dominated by the hard pomeron can be incorporated in the current scenario if the partial em gauge symmetry breaking for a nucleon is assumed .
this is due to the fact that like a lepton , the strange and charm quarks contain no nucleon number .
they do not couple to the goldstone boson from em @xmath13 symmetry breaking .
the effective em coupling of strange and charm quark to the up and down quark sector contains a strong interaction component
the current experimental data on the @xmath342 collision production of mesons seems to indicate @xcite that the soft pomeron transforms as a vector that couples to the a non - conserving current .
this is a natural consequences of the scenario proposed here if one assumes that there is a spontaneous em gauge symmetry partial breaking phase for a nucleon .
this is because if the soft pomeron behaves like a photon , then only the non - conserving core component of the em charge current in the color superconducting phase contribute to the observables , as it is discussed above .
the scenario proposed in this work depends on the assumption that there is a nucleon long before the experimental processes take place . for those @xmath343 creation processes like in the @xmath344 collision in the vacuum , the time for the initial @xmath345 creation is expected to be much faster than the collective multi - particle processes to let the excited system to cool to its ground state containing the observed @xmath343 pairs , in which the possible superconducting phase is formed .
so , the proposed hard pomeron behavior should not appear @xcite .
this is indeed observed behavior @xcite .
the possibility to search for a close - by metastable color superconducting phase for the strong interaction vacuum state in the presence of a nucleon is studied .
it is shown that the metastable color superconducting phase of the strong interaction vacuum state can manifest itself in high energy semi - leptonic neutral current electroweak interaction processes involving a nucleon through a mechanism of `` spontaneous partial breaking '' of the em @xmath346 gauge symmetry that is independent of any mass scale . in order to achieve this
, the physical processes in a system in which the em gauge symmetry is spontaneously partial broken leading to color superconductivity is studied model independently .
it is found that due the fact that the electric charge of quarks is fractional of the charge of a proton and is flavor dependent , the spontaneous partial breaking of the em gauge symmetry in the hadronic sector caused by a diquark condensation also breaks the global baryon number conservation .
the em gauge symmetry is only spontaneously partial broken in such a case where the goldstone bosons corresponding to the global symmetry breaking is not unphysical states in all channels of the reaction .
it is also shown that for a system in which the em gauge symmetry is spontaneously partial broken , due to the medium effects , the final phase space in a high energy semi - leptonic process contains an extra subspace compared to the allowed one due to the froissart bound for the high energy nucleon
nucleon scattering .
such an extra subspace can be considered as spanned by the ( lack of ) goldstone bosons of the the spontaneous global symmetry breaking which induces the spontaneous partial breaking of the em gauge symmetry .
it is found here that to explain some of the puzzling behavior of the nucleon structure functions at small @xmath0 , the violation of the gdh sum rule , the meson production in exclusive semi - leptonic dis and the current non - conservation in the meson production in high energy proton
proton collisions , the assumption that a nucleon is superconducting needs to be made .
the spin structure functions of the nucleons at small @xmath0 is perhaps the most sensitive ones to study the spontaneous partial breaking of em gauge symmetry inside a nucleon due to the presence of a very singular component in the experimental observables @xmath347 and @xmath348 , which was seen in the most recent data at @xmath0 as low as @xmath325 for @xmath349 .
the magnitude of the effects of the spontaneous partial breaking of the em gauge symmetry observed in the em properties of a nucleon at high energies , if proven true , means that there is at least one color superconducting metastable phase for the hadronic vacuum state with low enough energy density that can be turned into the stable one at relatively low density ( compared to the nuclear saturation density ) rather than at high densities . because of the simplicity of the single nucleon system , the signal for the possible color superconducting phase of the vacuum state is relatively clear to allow a further more detailed experimental study in the existing and planed facilities measuring the high energy em and neutral current weak responses of a nucleon at small @xmath7 and at small @xmath0 .
it must be emphasized that the results of this paper and earlier related ones , which concern with the virtual phases of the vacuum state , does neither uniquely imply nor disfavor the popular diquark model or quark quark clustering model for a nucleon .
these models are based on hypothetical binding / clustering between a pair of valence quarks inside a nucleon .
such a binding / clustering at valence level has no direct relation to the conclusion draw here .
in fact a model for a nucleon can be build without any quark
quark clustering even in the presence of a virtual color superconducting phase for the vacuum @xcite .
this work is supported by the national natural science foundation of china under contract 19875009 and a research fund from the department of education of china .
d. bailin and a. love , phys
. rept . * 107 * ( 1984 ) , 325 .
d. t. son , phys . rev .
* d59 * ( 1999 ) , 094019 .
s. d. h. hsu , m. schwetz , report no .
hep - ph/9908310 .
s. ying , phys . lett . *
b283 * ( 1992 ) , 341 .
s. ying , ann .
( n.y . ) * 250 * ( 1996 ) , 69 .
s. ying , report no .
hep - ph/9604255 .
s. ying , ann .
( n.y . ) * 266 * ( 1998 ) , 295 .
s. ying , report no .
hep - ph/0003001 .
d. diakonov , h. forkel and m. lutz , phys . lett .
* b373 * ( 1996 ) , 147 .
f. wilczek , report no .
hep - ph/9906395 .
e. v. shuryak , report no .
hep - ph/9908290 .
ying , report no .
hep - ph/9704216 .
s. ying , report no .
hep - th/9908087 .
golzman and k. varga , report no .
hep - ph/9901439 .
s. ying , report no .
hep - ph/0003050 .
s. ying , report no .
nucl - th/9411017 .
s. ying , commun .
theor . phys .
* 28 * ( 1997 ) , 301 .
s. ying , j. phys .
* g22 * ( 1996 ) , 293 .
h1 collab .
, s. aid et al .
, nucl . phys . *
b470 * ( 1996 ) , 3 ; c. adloff et al . , preprint desy 9702 .
zeus collab .
, m. derrick et al .
, z. phys .
* c72 * ( 1996 ) , 399 ; j. breitweg et al . ,
preprint desy 97 - 135 .
e665 collab . ,
phys . rev .
* d54 * ( 1996 ) , 3006 .
e143 collab . , k. abe et al . ,
phys . rev .
* d58 * ( 1998 ) , 112003 .
smc collab . ,
b. adeva et al . , phys
* d60 * ( 1999 ) , 072004 .
e143 collab . , k. abe et al .
* 79 * ( 1997 ) , 26 ; phys . lett .
* b404 * ( 1997 ) , 377 ; phys . lett .
* 405 * ( 1997 ) , 180 .
e866/nusea collab .
, e. a. hawker et al .
* 80 * ( 1998 ) , 3715 . c. boros , j. t. londergan , and a. w. thomas , phys .
* 81 * ( 1998 ) , 4075 ; phys . rev .
* d59 * ( 1999 ) , 074021 . c. boros , j. t. londergan , and a. w. thomas ,
report no .
hep - ph/9908280 .
o. j. p. eboli , e. m. gregores , and f. halzen , phys
. rev . * d58 * ( 1998 ) , 114005 .
p. v. landshoff , report no .
hep - ph/9905230 .
r. d. ball and p. v. landshoff , report no .
hep - ph/9912445 .
a. donnachie and p. v. landshoff , phys .
* b437 * ( 1998 ) , 408 .
h. abramowicz , e. m. levin , a. levy , and u. maor , phys .
lett . * b269 * ( 1991 ) , 465 .
a. j. askow , j. kwiecinski , a. d. martin , and p. j. sutton , phys . rev .
* d49 * ( 1994 ) , 4402 ; n. n. nikolaev , b. g. zakharov , and v. r. zoller , jept letter , * 59 * ( 1994 ) , 6 ; a. d. martin w. j. stirling , and r. g. roberts , phys . rev .
* d47 * ( 1993 ) , 867 ; j. c. collins and p. v. landshoff , phys . lett .
* 276 * ( 1992 ) , 196 .
g. f. bertsch , l. frankfurt , and m. strikman , science ( 1993 ) , 773 .
w. c. leung et al .
, phys . lett . *
b317 * ( 1993 ) , 655 ; p.
z. quintas et al .
rev . lett .
* 71 * ( 1993 ) , 1307 .
f. e. close , g. a. schuler , phys
b464 * ( 1999 ) , 279 ; the wa102 collaboration : d. barberis , et al , report no .
hep - ex/9909013 .
i. p. ivanov , n. n. nikolaev , a. v. pronyaev and w. schfer , report no .
hep - ph/9903228 . ,
e. farhi and r. jackiw , eds . ,
world scientific , singapore , 1982
. m. alford , j. berges , and k. rajagopal , nucl .
* b571 * ( 2000 ) , 269 . g. w. carter and d. diakonov , report no .
hep - ph/0001318 .
d. drechsel , g. g. krein , phys .
* d58 * ( 1998 ) , 116009 .
d. drechsel , s. s. kamalov , g. krein , and l. tiator , phys . rev .
* d59 * ( 1999 ) , 094021 .
s. okubo , phys . lett .
* 5 * ( 1963 ) , 165 ; g. zweig , cern report , no .
8419 ( 1964 ) , unpublished ; i. iuzuka , prog .
* 37 - 38 * ( 1966 ) , 21 .
r. s. thorne , phys .
b418 * ( 1998 ) , 371 .
a. d. martin et al . , eur .
j. * c4 * ( 1998 ) , 463 .
m. dueren , report no .
hep - ex/9905053 .
a. donnarchie and p. v. landshoff , nucl
* b244 * ( 1984 ) , 322 .
the zeus collaboration , j. breitweg et al . , z. phys .
* c75 * ( 1997 ) , 215 ; the h1 collaboration , s. aid et al . , nucl .
phys . * b472 * ( 1996 ) , 3 .
a. donnarchie and p. v. landshoff , report no .
hep - ph/9912312 .
s. soeldner - rembold , report no .
hep - ex/9805008 .
s. ying , report no .
hep - ph/0006111 . | a collection of the physical observables , related to the electromagnetic properties of a nucleon , to investigate the non perturbative quantum fluctuations in the strong interaction vacuum state under the influence of at least one close by ( in energy density ) color superconducting phase found in several qcd motivated model calculations , are studied .
it is shown that the spontaneous breaking of the electromagnetic gauge symmetry in the color superconducting phase of strong interaction can result in relatively clean signals in high energy processes , especially in the semi - leptonic deep inelastic scattering ones , due to a kind of electromagnetic induced strong interaction .
a new type of mechanism , which is a generalization of the higgs one , through which the local electromagnetic gauge symmetry is spontaneously broken by a spontaneous breaking of the global baryon ( nucleon ) number conservation , is revealed . a model independent assessment of the question of how far is the color superconducting phase of the strong interaction from its vacuum phase is made by studying currently available experimental data on the electromagnetic responses of a nucleon at high energies .
it is shown that based on our current knowledge about a nucleon , it is quite likely that there is at least one color superconducting phase for the strong interaction that is close enough to the vacuum state so that its effects can even be seen in high energy processes besides heavy ion collisions .
# 1 # 1 # 1#2| |
we are grateful to jan vorberger for providing the montroll - ward and e@xmath98 data shown in fig .
[ intro_pic ] .
this work was supported by the deutsche forschungsgemeinschaft via project bo1366 - 10 and via sfb tr-24 project a9 as well as grant shp00015 for cpu time at the norddeutscher verbund fr hoch- und hchstleistungsrechnen ( hlrn ) .
ts acknowledges the support of the us doe / nnsa under contract no .
de - ac52 - 06na25396 .
fdm is funded by an imperial college phd scholarship .
fdm and wmcf used computing facilities provided by the high performance computing service of imperial college london , by the swiss national supercomputing centre ( cscs ) under project i d s523 , and by archer , the uk national supercomputing service , under epsrc grant ep / k038141/1 and via a rap award .
fdm and wmcf acknowledge the research environment provided by the thomas young centre under grant no .
tyc-101 .
perdew and a. zunger , self - interaction correction to density - functional approximations for many - electron systems , http://link.aps.org/doi/10.1103/physrevb.23.5048 [ _ phys .
b _ * 23 * , 5048 ] ( 1981 ) m.d .
knudson _ et al .
_ , probing the interiors of the ice giants : shock compression of water to 700 gpa and @xmath99 , http://link.aps.org/doi/10.1103/physrevlett.108.091102 [ _ phys .
lett . _ * 108 * , 091102 ] ( 2012 ) p.f .
schmit _ et al .
_ , understanding fuel magnetization and mix using secondary nuclear reactions in magneto - inertial fusion http://link.aps.org/doi/10.1103/physrevlett.113.155004 [ _ phys .
lett . _ * 113 * , 155004 ] ( 2014 ) o.a .
hurricane _ et al .
_ , inertially confined fusion plasmas dominated by alpha - particle self - heating , http://www.nature.com/nphys/journal/vaop/ncurrent/full/nphys3720.html [ _ nature phys .
_ 3720 ] ( 2016 ) k. burke , j.c .
smith , p.e .
grabowski , and a. pribram - jones , exact conditions on the temperature dependence of density functionals , http://link.aps.org/doi/10.1103/physrevb.93.195132 [ _ phys .
b _ * 93 * , 195132 ] ( 2016 ) v.v .
karasiev , l. calderin , and s.b .
trickey , the importance of finite - temperature exchange - correlation for warm dense matter calculations , http://link.aps.org/doi/10.1103/physreve.93.063207 [ _ phys .
e _ * 93 * , 063207 ] ( 2016 ) a. pribram - jones , p.e .
grabowski , and k. burke , thermal density functional theory : time - dependent linear response and approximate functionals from the fluctuation - dissipation theorem , http://link.aps.org/doi/10.1103/physrevlett.116.233001 [ _ phys .
_ * 116 * , 233001 ] ( 2016 ) u. gupta and a.k .
rajagopal , exchange - correlation potential for inhomogeneous electron systems at finite temperatures , http://link.aps.org/doi/10.1103/physreva.22.2792 [ _ phys .
a _ * 22 * , 2792 ] ( 1980 ) f. perrot and m.w.c . dharma - wardana , exchange and correlation potentials for electron - ion systems at finite temperatures , http://link.aps.org/doi/10.1103/physreva.30.2619 [ _ phys
. rev . a _
* 30 * , 2619 ] ( 1984 ) s. tanaka and s. ichimaru , thermodynamics and correlational properties of finite - temperature electron liquids in the singwi - tosi - land - sjlander approximation , http://journals.jps.jp/doi/abs/10.1143/jpsj.55.2278 [ _ j. phys
. soc .
_ * 55 * , 2278 - 2289 ] ( 1986 ) m.w.c
. dharma - wardana and f. perrot , simple classical mapping of the spin - polarized quantum electron gas : distribution functions and local - field corrections , http://link.aps.org/doi/10.1103/physrevlett.84.959 [ _ phys .
_ * 84 * , 959 - 962 ] ( 2000 ) e.y .
loh , j.e .
gubernatis , r.t .
scalettar , s.r .
white , d.j .
scalapino and r.l .
sugar , sign problem in the numerical simulation of many - electron systems , http://link.aps.org/doi/10.1103/physrevb.41.9301[_phys .
b _ * 41 * , 9301 - 9307 ] ( 1990 ) m. troyer and u.j
. wiese , computational complexity and fundamental limitations to fermionic quantum monte carlo simulations , http://link.aps.org/doi/10.1103/physrevlett.94.170201[_phys .
lett . _ * 94 * , 170201 ] ( 2005 ) e.w .
brown , b.k .
clark , j.l .
dubois and d.m .
ceperley , path - integral monte carlo simulation of the warm dense homogeneous electron gas , http://link.aps.org/doi/10.1103/physrevlett.110.146405 [ _ phys .
lett . _ * 110 * , 146405 ] ( 2013 ) s. groth , t. schoof , t. dornheim , and m. bonitz , _ ab initio _ quantum monte carlo simulations of the uniform electron gas without fixed nodes , http://link.aps.org/doi/10.1103/physrevb.93.085102 [ _ phys .
b _ * 93 * , 085102 ] ( 2016 ) t. dornheim , s. groth , t. schoof , c. hann , and m. bonitz , _ ab initio _ quantum monte carlo simulations of the uniform electron gas without fixed nodes : the unpolarized case , http://link.aps.org/doi/10.1103/physrevb.93.205134 [ _ phys .
b _ * 93 * , 205134 ] ( 2016 ) t. schoof , m. bonitz , a.v .
filinov , d. hochstuhl and j.w .
dufty , configuration path integral monte carlo , http://onlinelibrary.wiley.com/doi/10.1002/ctpp.201100012/abstract [ _ contrib .
plasma phys .
_ * 51 * , 687 - 697 ] ( 2011 ) t. schoof , s. groth and m. bonitz , towards ab initio thermodynamics of the electron gas at strong degeneracy , http://onlinelibrary.wiley.com/doi/10.1002/ctpp.201400072/abstract [ _ contrib .
plasma phys .
_ * 55 * , 136 - 143 ] ( 2015 ) t. schoof , s. groth , j. vorberger and m. bonitz , _ ab initio _ thermodynamic results for the degenerate electron gas at finite temperature , http://link.aps.org/doi/10.1103/physrevlett.115.130402 [ _ phys .
_ * 115 * , 130402 ] ( 2015 ) t. dornheim , s. groth , a. filinov and m. bonitz , permutation blocking path integral monte carlo : a highly efficient approach to the simulation of strongly degenerate non - ideal fermions , http://iopscience.iop.org/1367-2630/17/7/073017 [ _ new j. phys .
_ * 17 * , 073017 ] ( 2015 ) t. dornheim , t. schoof , s. groth , a. filinov , and m. bonitz , permutation blocking path integral monte carlo approach to the uniform electron gas at finite temperature , http://scitation.aip.org/content/aip/journal/jcp/143/20/10.1063/1.4936145 [ _ j. chem
. phys . _ * 143 * , 204101 ] ( 2015 ) n.s . blunt , t.w .
rogers , j.s .
spencer and w.m .
foulkes , density - matrix quantum monte carlo method , http://link.aps.org/doi/10.1103/physrevb.89.245124 [ _ phys .
b _ * 89 * , 245124 ] ( 2014 ) f.d .
malone _ et al .
_ , interaction picture density matrix quantum monte carlo , http://scitation.aip.org/content/aip/journal/jcp/143/4/10.1063/1.4927434 [ _ j. chem .
_ * 143 * , 044116 ] ( 2015 ) l.m .
fraser _ et al .
_ , finite - size effects and coulomb interactions in quantum monte carlo calculations for homogeneous systems with periodic boundary conditions , http://link.aps.org/doi/10.1103/physrevb.53.1814 [ _ phys . rev .
b _ * 53 * , 1814 ] ( 1996 ) n.d .
drummond , r.j .
needs , a. sorouri and w.m.c .
foulkes , finite - size errors in continuum quantum monte carlo calculations , http://link.aps.org/doi/10.1103/physrevb.78.125106 [ _ phys .
b _ * 78 * , 125106 ] ( 2008 ) s. chiesa , d.m .
ceperley , r.m .
martin , and m. holzmann , finite - size error in many - body simulations with long - range interactions , http://link.aps.org/doi/10.1103/physrevlett.97.076404 [ _ phys .
lett . _ * 97 * , 076404 ] ( 2006 ) c. lin , f.h .
zong and d.m .
ceperley , twist - averaged boundary conditions in continuum quantum monte carlo algorithms , http://link.aps.org/doi/10.1103/physreve.64.016702[_phys .
e _ * 64 * , 016702 ] ( 2001 ) h. kwee , s. zhang , and h. krakauer , finite - size correction in many - body electronic structure calculations , http://link.aps.org/doi/10.1103/physrevlett.100.126404 [ _ phys .
_ * 100 * , 126404 ] ( 2008 ) in the considered temperature range shell - filling effects are negligible and twist - averaging @xcite is not required .
note that @xmath1 is sufficient to obtain @xmath2 which gives access to all other quantities , including the kinetic energy .
we ( i ) perform a linear extrapolation and ( ii ) average over the last few data points where @xmath104 are indistinguishable within the statistical error bars .
our final result for @xmath104 is the mean of ( i ) and ( ii ) , which constitute reasonable lower and upper bounds , respectively .
the discretization fsc is insensitive to the choice of sf because the discretization error is unaffected by constant shifts in the model sf ( the three sfs in the inset of fig .
[ main_sk ] are nearly parallel ) . to obtain the bcdc data points for @xmath19 in fig .
[ pot ] , we subtracted from the rpimc results for @xmath1 of the supplement of ref .
@xcite the tabulated fscs and added the correct bcdc - fsc , eq .
( [ eq_bcdc ] ) , see ref .
@xcite for details .
karasiev , t. sjostrom , j. dufty and s.b .
trickey , accurate homogeneous electron gas exchange - correlation free energy for local spin - density calculations , http://link.aps.org/doi/10.1103/physrevlett.112.076403 [ _ phys .
_ * 112 * , 076403 ] ( 2014 ) n. crouseilles , p .- a .
hervieux , g. manfredi , quantum hydrodynamic model for the nonlinear electron dynamics in thin metal films , http://link.aps.org/doi/10.1103/physrevb.78.155412 [ _ phys .
b _ * 78 * , 155412 ] ( 2008 ) d. michta , f. graziani , and m. bonitz , quantum hydrodynamics for plasmas a thomas - fermi theory perspective , http://onlinelibrary.wiley.com/doi/10.1002/ctpp.201500024/abstract [ _ contrib .
plasma phys .
_ * 55 * , 437 ] ( 2015 ) k.p . driver , and b. militzer , all - electron path integral monte carlo simulations of warm dense matter : application to water and carbon plasmas , http://link.aps.org/doi/10.1103/physrevlett.108.115502 [ _ phys . rev .
_ * 108 * , 115502 ] ( 2012 ) b. militzer , and k.p .
driver , development of path integral monte carlo simulations with localized nodal surfaces for second - row elements , http://link.aps.org/doi/10.1103/physrevlett.115.176403 [ _ phys .
_ * 115 * , 76403 ] ( 2015 ) | we perform _ ab initio _ quantum monte carlo ( qmc ) simulations of the warm dense uniform electron gas in the thermodynamic limit . by combining qmc data with linear response theory we are able to remove finite - size errors from the potential energy over the entire warm dense regime , overcoming the deficiencies of the existing finite - size corrections by brown _
et al . _ [ prl * 110 * , 146405 ( 2013 ) ] .
extensive new qmc results for up to @xmath0 electrons enable us to compute the potential energy @xmath1 and the exchange - correlation free energy @xmath2 of the macroscopic electron gas with an unprecedented accuracy of @xmath3 . a comparison of our new data to the recent parametrization of @xmath2 by karasiev _
et al . _ [ prl * 112 * , 076403 ( 2014 ) ] reveals significant inaccuracies of the latter .
the uniform electron gas ( ueg ) , consisting of electrons on a uniform neutralizing background , is one of the most important model systems in physics . besides being a simple model for metals
, the ueg has been central to the development of linear response theory and more sophisticated perturbative treatments of solids , the formulation of the concepts of quasiparticles and elementary excitations , and the remarkable successes of density functional theory . the practical application of ground - state density functional theory in condensed matter physics , chemistry and materials science rests on a reliable parametrization of the exchange - correlation energy of the ueg @xcite , which in turn is based on accurate quantum monte carlo ( qmc ) simulation data @xcite . however , the charged quantum matter in astrophysical systems such as planet cores and white dwarf atmospheres @xcite is at temperatures way above the ground state , as are inertial confinement fusion targets @xcite , laser - excited solids @xcite , and pressure induced modifications of solids , such as insulator - metal transitions @xcite .
this unusual regime , in which strong ionic correlations coexist with electronic quantum effects and partial ionization , has been termed `` warm dense matter '' and is one of the most active frontiers in plasma physics and materials science .
the warm dense regime is characterized by the existence of two comparable length scales and two comparable energy scales .
the length scales are the mean interparticle distance , @xmath4 , and the bohr radius , @xmath5 ; the energy scales are the thermal energy , @xmath6 , and the electronic fermi energy , @xmath7 . the dimensionless parameters @xcite @xmath8 and @xmath9 are of order unity .
because @xmath10 , the use of ground - state density functional theory is inappropriate and extensions to finite @xmath11 are indispensible ; these require accurate exchange - correlation functionals for finite temperatures @xcite . because neither @xmath12 nor @xmath13 is small , there are no small parameters , and weak - coupling expansions beyond hartree - fock potential energy per particle of the unpolarized ueg at @xmath14 and @xmath15 .
the exact cpimc results for different system sizes are indicated by green crosses ; the yellow asterisks show these results after the @xmath16 finite - size correction from eq .
( [ eq_bcdc ] ) has been applied .
the horizontal arrows show the results of various many - body theories ( rpa , stls @xcite , montroll - ward [ mw ] , and @xmath17 [ e4 ] @xcite ; see text ) .
the black lines are two different , equally plausbile , extrapolations of the qmc data to infinite system size @xcite .
, scaledwidth=48.0% ] such as the montroll - ward ( mw ) and @xmath17 ( e4 ) approximations @xcite , as well as linear response theory within the random - phase approximation ( rpa ) break down @xcite .
finite-@xmath11 singwi - tosi - land - sjlander ( stls ) @xcite local - field corrections allow for an extension to moderate coupling @xcite , but exhibit non - physical behavior at short distances for moderate to low densities .
further , quantum - classical mapping @xcite allows for semi - quantitative descriptions of warm dense matter in limiting cases .
therefore , an accurate description of warm dense matter in general and of the warm dense ueg in particular can only be achieved using computational approaches , primarily quantum monte - carlo ( qmc ) methods which , however , are hampered by the fermion sign problem @xcite . the pioneering qmc simulations of the warm dense ueg by brown _
et al . _
@xcite eliminated the sign problem by invoking the ( uncontrolled ) fixed - node approximation @xcite , but were nevertheless restricted to small systems of @xmath18 ( spin - polarized ) and @xmath19 ( unpolarized ) electrons and to moderate densities , @xmath20 .
recently , we were able to show @xcite that accurate simulations of these systems are possible over a broad parameter range without any nodal restriction .
our approach combines two independent methods , configuration path - integral monte carlo ( cpimc ) @xcite and permutation blocking pimc @xcite , which allow for accurate simulations at high ( @xmath21 ) and moderate densities ( @xmath22 and @xmath23 ) , respectively .
an independently developed third approach , density matrix qmc @xcite , confirmed the excellent quality of these results .
the only significant errors remaining are finite - size effects @xcite , which arise from the difference between the small systems simulated and the infinite [ thermodynamic limit ( tdl ) ] system of interest .
direct extrapolation to the tdl @xcite is extremeley costly and also unreliable unless the form of the function to be extrapolated is known ; the two black lines in fig . [ intro_pic ] show two equally reasonable extrapolations @xcite that reach different limits .
furthermore , the parameter - free finite - size correction ( fsc ) proposed in ref .
@xcite ( see eq .
( [ eq_bcdc ] ) below ) turns out to be inappropriate in parts of the warm dense regime .
the problem is clear from inspection of the yellow asterisks in fig .
[ intro_pic ] , which include this fsc but remain system - size dependent . in this letter , we close the gap between the finite-@xmath24 qmc data and the tdl by deriving a highly accurate fsc for the interaction energy .
this allows us to obtain precise ( on the level of @xmath25 ) results for the exchange - correlation free energy , making possible the _ ab initio _ computation of arbitrary thermodynamic quantities over the entire warm dense regime .
* theory*. consider a finite unpolarized ueg of @xmath24 electrons subject to periodic boundary conditions .
the hamiltonian is @xmath26 , where @xmath27 is the kinetic energy of the @xmath24 electrons in the cell and @xmath28 is the coulomb interaction energy per unit cell of an infinite periodic array of images of that cell . the ewald pair potential @xmath29 and madelung constant @xmath30 are defined in refs . @xcite .
we use hartree atomic units throughout this work .
the expected value of @xmath31 carries a finite - size error @xcite that is the difference between the potential energy @xmath32 per electron in the infinite system and its value @xmath33 in the finite system .
this difference may be expressed in terms of the static structure factor ( sf ) as follows : @xmath34}{n } = \underbrace { \frac{1}{2}\int_{k<\infty}\frac{\textnormal{d}\mathbf{k}}{(2\pi)^3 } \left[s(k)-1\right]\frac{4\pi}{k^2 } } _ { v } \notag \\ \label{delta_v_1 } & \hspace*{5em } - \underbrace { \left ( \frac{1}{2l^3}\sum_{\mathbf{g}\ne\mathbf{0}}\left[s_n(\mathbf{g})-1\right]\frac{4\pi}{g^2}+\xi_\textnormal{m}\right ) } _ { v_n / n } , \end{aligned}\ ] ] where @xmath35 and @xmath36 are , respectively , the length and reciprocal lattice vector of the simulation cell , @xmath37 [ @xmath38 is the sf of the infinite [ finite ] system . a first source of fs error in eq .
( [ delta_v_1 ] ) is the replacement of @xmath37 in the first term by its finite - size analogue @xmath39 in the second term .
however , this effect is negligible , as we will demonstrate in fig .
[ main_sk ] .
thus the main source of error is the discretization of the integral in the first term to obtain the sum in the second term .
in fact , chiesa _ et al . _
@xcite suggested that the main contribution to eq .
( [ delta_v_1 ] ) comes from the omission of the @xmath40 term from the summation @xcite .
as is well known , the rpa becomes exact in the limit of small @xmath41 , and the expansion of @xmath37 around @xmath42 at finite @xmath11 is given by @xcite @xmath43 where @xmath44 is the inverse temperature and @xmath45 is the plasma frequency . the finite-@xmath11 version of the chiesa fsc @xcite
, @xmath46 would be sufficient if : ( i ) @xmath47 were accurate for the smallest nonzero @xmath41 in the qmc simulation , @xmath48 ; and ( ii ) all contributions to eq .
( [ delta_v_1 ] ) not accounted for by the inclusion of the @xmath49 term were negligible .
as we demonstrate below , for high temperatures and intermediate to high densities , both conditions are strongly violated .
thus , we need to use an improved model sf , @xmath50 , to compute the discretization error , @xmath51 = \frac{\delta v_n \left[s_\textnormal{model}(k),s_\textnormal{model}(k)\right]}{n } , \label{eq_trial}\ ] ] in eq .
( [ delta_v_1 ] ) .
a natural strategy is to combine the qmc data for @xmath52 with an approximation that is accurate for all @xmath41 up to ( at least ) @xmath53 .
static structure factors for @xmath14 , @xmath54 and three values of @xmath24 . in panel
a ) , the discrete qmc @xmath41-points are plotted as vertical lines for @xmath55 , the minimum k - values for @xmath19 and @xmath56 are indicated by the green and yellow line .
the colored horizontal bars indicate the @xmath41-ranges where @xmath57 ( red ) , @xmath58 rpa ( grey ) and @xmath59 ( light blue ) are accurate .
panel b ) shows that the qmc results for @xmath37 converge rapidly with @xmath24 ( see the colored symbols in the inset ) .
the black curve shows @xmath60 connecting @xmath61 at small @xmath41 with the qmc data for @xmath55 which yields accurate results for all @xmath41 . , scaledwidth=48.0% ] * results*. in fig
. [ main_sk ] , we analyze the static sf for @xmath14 and a comparatively high density case , @xmath15 , for three different particle numbers . the use of a finite simulation cell subject to periodic boundary conditions discretizes the momentum , so qmc data are available only at the discrete @xmath41-points indicated by the vertical lines in the top panel . as shown in the inset , the qmc @xmath37 is well converged with respect to system size for surprisingly small @xmath24 , providing justification to set @xmath62 .
therefore , the fs error of @xmath33 reduces as @xmath24 increases primarily because the @xmath41-grid becomes finer and @xmath53 decreases .
the figure also allows us to study the performance of the three analytical structure factors , @xmath58 , @xmath57 @xcite and @xmath59 .
we clearly observe that @xmath47 is only accurate for @xmath63 , explaining why the bcdc fsc , eq .
( [ eq_bcdc ] ) , fails .
in contrast , @xmath64 and @xmath61 match the qmc data much better .
on the left - hand side of panel a ) , we indicate the @xmath41-ranges over which the three models are accurate , showing that only @xmath61 connects smoothly to the qmc data . at larger @xmath41 , @xmath58 and @xmath57 exhibit significant deviations from the qmc data , although stls is more accurate . for completeness , we mention that , when the density is lowered , the @xmath65ranges of accurate behavior of @xmath58 , @xmath57 and @xmath59 continuously increase @xcite . for example , at @xmath66 , both @xmath58 and @xmath57 smoothly connect to the qmc data whereas for @xmath67 this
is observed even for @xmath47 revealing that there the bcdc fsc is accurate .
based on this behavior , an obvious way to construct a model sf that is accurate over the entire @xmath41-range for all warm dense matter parameters is to combine the qmc data with the stls data at small @xmath41 . the result is denoted @xmath68 and computed via a spline function .
the excellent behavior is illustrated by the black line in panel b ) of fig .
[ main_sk ] and in the inset .
this quasi - exact sf is the proper input to compute the discretization error from eq .
( [ eq_trial ] ) . *
a ) * finite - size corrected qmc data for the potential energy for @xmath14 and @xmath15 .
the yellow asterisks are obtained using eq .
( [ eq_bcdc ] ) ; the red diamonds use the combined sf @xmath69 ( cf . fig .
[ main_sk ] ) to evaluate the discretization error , eq .
( [ eq_trial ] ) .
* b ) * magnified part of panel a ) including an extrapolation of the residual finite - size error to the tdl ( the red cross ) .
results obtained using only the full rpa ( blue ) and stls structure factors ( black ) in eq .
( [ eq_trial ] ) are also shown .
, scaledwidth=45.0% ] the results of this procedure are shown in fig . [ rs05 ] for the most challenging high - density case , @xmath15 and @xmath14 .
clearly , the raw qmc data ( green crosses ) suffer from severe finite - size errors of order @xmath70 for system sizes from @xmath56 to @xmath71 .
these errors do not exhibit the @xmath72 behavior predicted by eq .
( [ eq_bcdc ] ) , and the bcdc - corrected qmc data ( yellow asterisks ) do not fall on a horizontal line .
in contrast , using @xmath73 $ ] produces results that are very well converged for all system sizes considered , including even @xmath56 ( red diamonds ) .
panel b ) of fig .
[ rs05 ] shows that the removal of the discretization error has reduced the fs bias by two orders of magnitude .
the residual error , @xmath74 , is due to the small finite - size effects in the qmc data for @xmath75 itself and exhibits a linear behavior in @xmath76 .
thus , it is possible to determine the potential energy in the tdl ( the red cross in the bottom panel ) with a reliable error bar @xcite . to further explore the properties of our discretization formula for the fs error , we recompute @xmath77 using the purely theoretical stls and rpa sfs as @xmath78 in eq .
( [ eq_trial ] ) .
the fs - corrected data are depicted by the black squares and blue circles in panel b ) of fig .
[ rs05 ] , respectively .
surprisingly , we find very good agreement with the fscs derived from the substantially more accurate @xmath68 .
hence , despite their significant deviations from the qmc data at intermediate @xmath41 ( cf .
inset in panel b of fig .
[ main_sk ] ) , @xmath61 and @xmath64 are sufficiently accurate to account for the discretization error of the potential energy @xcite . since @xmath68 is sensitive to statistical noise , computing the fsc
solely from @xmath61 or @xmath64 is in fact the preferred approach . of course , this unexpectedly simple solution to the finite - size - correction problem does not eliminate the need for accurate finite-@xmath24 qmc data , the quality of which sets the base line for our thermodynamic result , @xmath79 $ ] .
using instead the stls or rpa sf to estimate @xmath80 as well as @xmath77 poorly accounts for the short - range correlations and , even for @xmath14 and @xmath15 , leads to @xmath81 errors ( cf .
fig .
[ intro_pic ] ) , which further increase with @xmath12 .
potential energy of the ueg in the tdl . *
a ) * our new fs - corrected qmc data , the fits to our data ( see eq .
( s.2 ) of ref .
@xcite ) , and the rpimc results of brown _ et al . _
@xcite , which include bcdc fscs .
* b ) * relative deviations of our data ( for @xmath82 ) and brown s bcdc - corrected data from the corresponding fit .
* c ) * relative deviation of our exchange correlation free energies from the fit of ref .
@xcite for five temperatures .
for details see ref .
@xcite .
, scaledwidth=44.0% ] by performing extensive qmc simulations and applying our fsc to results for various system sizes @xmath24 to allow extrapolatation of the residual fs error , we obtain the potential energy of the ueg in the tdl over a very broad density range , @xmath83 .
the results are displayed in fig .
[ pot ] for five different temperatures and listed in a table in the supplement @xcite .
we also compare our results to the most accurate data previously available the rpimc results of brown _ et al . _ ( bcdc , circles ) , which were corrected using the bcdc fsc , eq .
( [ eq_bcdc ] ) @xcite .
we underline that these results were limited to moderate densities , @xmath84 , but even there substantially deviate from our data .
the error increases rapidly with density and temperature reaching @xmath85 for @xmath66 and @xmath86 @xcite .
finally , we obtain the exchange - correlation free energy from a fit to the potential energy , regarded as a function of @xmath12 for fixed @xmath87 .
panel b ) of fig .
[ pot ] shows that the functional form assumed ( eq .
( s.2 ) in ref .
@xcite ) is indeed appropriate as no systematic deviations between the qmc data and the fit ( red crosses , @xmath86 ) are observed . in panel c ) of fig .
[ pot ] , we compare our new data for @xmath2 to the recent parametrization by karasiev _
et al . _
@xcite . by design ,
both curves coincide in the limit @xmath88 , approaching the exact asymptotic value known from hartree - fock theory ( for @xmath89 ) .
while both results are in very good agreement for @xmath90 , we observe severe deviations of up to @xmath91 at @xmath86 [ @xmath92 at @xmath14 ] . despite the systematic rpimc bias and the lack of data for @xmath94 prior to our work ,
the major cause of the disagreement is the inadequacy of the bcdc fscs for high temperature and small @xmath12 .
the absolute data for @xmath2 are provided in ref .
@xcite .
* summary and discussion*. we have presented a simple but highly accurate procedure for removing finite - size errors from _ ab initio _ finite-@xmath24 qmc data for the potential energy @xmath1 of the ueg at finite temperature .
this is achieved by adding to the qmc results the discretization error @xmath95 $ ] , eq .
( [ eq_trial ] ) , computed using simple approximate structure factors based on the rpa or stls approximations .
our finite - size - corrected results include excellent descriptions of both the exchange and short - range correlation effects ( from the qmc data ) and the long - range correlations ( via the rpa or stls corrections ) .
these results constitute the first _ ab initio _ thermodynamic data for the warm dense electron gas free of the limitations of many - body approximations or systematic simulation biases such as fixed - node error . for temperatures above half the fermi temperature and a density range covering six orders of magnitude ( @xmath96 )
, we achieve an unprecedented accuracy of @xmath97 ; our results will therefore serve as valuable benchmarks for the development of accurate new theories and simulation schemes .
in particular , we observe that the recent results of brown _ et al .
_ @xcite , which were obtained by applying the bcdc fsc from eq .
( [ eq_bcdc ] ) to rpimc data , exhibit deviations of up to @xmath85 .
the recent parametrization of @xmath2 by karasiev _
et al . _
@xcite , which was mainly based on the data by brown _
et al . _ , uses a good functional form but exhibits errors of up to @xmath91 at high temperatures . in future work
, we will extend our database of results and use it to provide much more accurate parametrizations of @xmath2 with respect to density , temperature , and spin polarization .
these parameterizations will be a key input for finite - t density functional theory , which remains the method of choice for treating realistic multi - component warm dense matter , and for time - dependent approaches such as quantum hydrodynamics @xcite and time - dependent density functional theory .
finally , our fsc procedure is expected to be of value for other simulations of warm dense plasmas @xcite . |
the dynamics and roughening of moving interfaces in disordered media has been a subject of great interest in non - equilibrium statistical physics since the 1980s .
relevant examples of physically and technologically important processes include thin film deposition @xcite , fluid invasion in porous media @xcite , and wetting and propagation of contact lines between phase boundaries @xcite .
the understanding of the underlying physics involved in interface roughening is crucial to the control and optimization of these processes .
significant progress in the theoretical study of interface dynamics has been made and a number of theories have been developed @xcite , which in some selected cases agree well with the experimental findings @xcite .
most of the theoretical understanding in this field is based on modeling interface roughening with a local stochastic equation of motion for the single - valued height variable of the interface .
however , there are several cases of interest where such an approach can not be justified _
e.g. _ due to conservation laws in the bulk .
this is especially true for processes such as fluid invasion in porous media , which is often experimentally studied by hele - shaw cells @xcite or imbibition of paper @xcite .
it has been shown that in such cases spatially local theories can not provide a complete description of the underlying dynamics .
for describing the diffusive invasion dynamics in such systems , a phase - field model explicitly including the local liquid bulk mass conservation law has been developed and applied to the dynamics of 1d imbibition fronts in paper @xcite .
this was achieved by a generalized cahn - hilliard equation with suitable boundary conditions , which couple the system to the reservoir .
numerical results for roughening from the model are in good agreement with relevant experiments @xcite .
one of the great advantages of the phase - field approach is that it s possible to analytically derive equations of motion for the phase boundaries in the so - called sharp interface limit @xcite .
most recently , we have developed a systematic formalism to derive such equations for the 2d meniscus and 1d contact line dynamics of fluids in capillary rise @xcite .
the equations are derived from the 3d bulk phase - field formulation , using variational approach as applied to relevant rayleigh dissipation and free energy functionals . through successive projections , equations of motion for
the 2d meniscus and 1d contact line can be derived .
the leading terms of such equations ( for small amplitude , long wavelength fluctuations ) can be shown to agree with results obtained from the sharp interface equations in the appropriate limits @xcite .
in addition to the need for non - local models to account for mass conservation , in hele - shaw and imbibition type of problems the inherent quenched disorder should be properly taken care of .
unlike thermal disorder , which is relatively easy to handle , quenched disorder depends on the height of the 1d interface @xmath0 as @xmath1 .
this makes its influence on interface roughening highly nontrivial , often leading to anomalous scaling @xcite .
currently , for such cases good agreement between theory , simulations and experiments has not been achieved . even on the experimental side some results such as the quantitative values of the scaling exponents , are not consistent and difficult to interpret .
very recently , soriano _ et al . _
@xcite conducted an experimental study of forced fluid invasion in a specially designed hele - shaw cell . the quenched disorder pattern in hele - shaw cell is realized by creating large number of copper islands that randomly occupy the sites of a square grid on a fiberglass substrate fixed on the bottom hele - shaw cell .
three different disorder patterns were used .
two of them are obtained by random selection of the sites of a square lattice .
the third kind of disorder is formed by parallel tracks , continuous in the interface growth direction and randomly distributed along the lateral direction .
it was found that for forced flow the temporal growth exponent @xmath2 which is nearly independent of experimental parameters and disorder patterns .
however , the spatial roughness exponent @xmath3 was found to be sensitive to experimental parameters and disorder patterns .
anomalous scaling with @xmath4 , and a local roughness exponent @xmath5 was found in disorder pattern with parallel tracks along the growth direction .
it was also demonstrated that such anomalous scaling is a consequence of different local velocities on the tracks and the coupling in the motion between neighboring tracks . on the theoretical side , for non - local hele - shaw and paper imbibition problems
there are two different approaches within the phase - field models to include additive quenched disorder .
et al . _
@xcite put the quenched disorder inside the chemical potential , the gradient of which is the driving force for interface motion . on the other hand , hernandez - machado _ et al . _
@xcite accounted for the effect of fluctuation of hele - shaw gap thickness as a mobility with quenched disorder in the phase - field model , while keeping the chemical potential free of noise .
these two approaches lead to quantitatively different roughening properties .
when considering the problem of capillary rise in a typical hele - shaw cell set - up between two rough walls more microscopically , the location of the surface of such corrugated walls in the cartesian coordinate system is a spatially fluctuating quantity , which indicates the presence of quenched disorder .
an experimental realization is given in @xcite . to treat this problem faithfully , in solving the phase - field equation
such a fluctuating wall surface should be treated as a physical boundary without phenomenologically adding quenched noise to the equation of motion as done previously .
consequently it is evident that a rigorous analytic treatment of such a problem is overwhelmingly difficult .
however , in this paper we demonstrate that with proper mathematical formulation of the problem , it is possible - albeit with some approximations - to analytically derive equations of motion for the meniscus and contact line dynamics .
most importantly , these equations incorporate the wall disorder in a natural way . to achieve this
, we utilize an explicit curvilinear coordinate transformation of the phase - field equation in order to apply projection methods to unravel the relevant physics in the limit of small disorder . to some extent ,
this kind of curvilinear coordinate transformation is similar to the boundary - fitted coordinate system frequently used in computational fluid dynamics ( cfd ) @xcite .
the outline of this paper is as follows : in chapter [ 2d ] we will consider the phenomenological 2d phase field model of capillary rise with quenched disorder in the mobility , similar to that in refs .
we will adapt the systematic projection method of kawasaki and ohta @xcite to obtain a linearized interface equation ( lie ) that describes small fluctuations of an interface , whose deterministic part reduces to the previous result @xcite in a special limit . to treat the problem rigorously , in chapter [ 3d ]
we will consider the full 3d phase field model with corrugated walls as the source of quenched disorder .
the transformation to curvilinear coordinates , as discussed above , is introduced to obtain linearized , effective bulk disorder from the original curvilinear boundary condition . following this we develop and apply a general projection scheme @xcite to obtain the effective equations of motion for small fluctuations of the 2d meniscus , and ultimately for the 1d contact line between the meniscus and the wall .
again , the deterministic parts of these equations reduce to previously known limits in special cases .
however , we demonstrate that the forms of the quenched noise terms derived here are different from the previous works .
a 2d phase - field model explicitly including the local conservation of bulk mass was introduced to study capillary rise by dub _
the bulk disorder in their model was included in the effective chemical potential . recently a similar model , where the disorder was considered through a stochastic mobility coefficient , was studied by hernandez - machado _ et al .
in particular , they assumed a one - sided mobility coefficient , which vanishes on one side of the interface . from this model they derived an equation of motion for small interface fluctuations . in this section , we will use the systematic projection method introduced by kawasaki and ohta @xcite , to derive the corresponding linearized interface equation ( lie ) describing small fluctuations in a sharp interface in a similar model . in our model
we assume the mobility to be independent of the phase , as in the previous works @xcite , but spatially stochastic , as in @xcite .
this corresponds to considering the invading fluid and the porous medium , but not the receding fluid .
this picture is valid when the receding fluid has low density and viscosity . in practice
this would mean a gas , such as air , being displaced by a liquid , such as water or oil .
the model allows a systematic projection of the effective noise term at the interface .
the phase field model describes capillary rise at a coarse - grained level with a phase field @xmath6 that obtains the value @xmath7 in the phase of the displaced fluid , and @xmath8 in the phase of the displacing fluid .
the phase field thus describes the effective component densities , and thus must be locally conserved .
an energy cost for an interface is included to obtain the free energy as @xmath9=\frac{1}{2}[\nabla\phi(\mathbf{x , t})]^2 + v(\phi(\mathbf{x , t})),\ ] ] where @xmath10 has two minima at @xmath8 and @xmath7 .
the details of @xmath10 are not relevant in the sharp interface limit , except to define the surface tension , so we can choose the standard ginzburg - landau form @xmath11 , where one of the phases can be set metastable by nonzero coefficient @xmath12 .
the equation of motion for the conserved phase field is given by the continuity equation @xmath13 and fick s law @xmath14 , where @xmath15 , and @xmath16 is the mobility that we choose to be a position dependent stochastic variable here .
the resulting equation of motion for the phase field is then given by @xmath17 = m\nabla\cdot ( 1+\xi(x))\nabla\left[v'(\phi)-\nabla^2\phi\right],\ ] ] where the variable @xmath18 is now the dimensionless , quenched noise .
the sharp interface limit of this model without the noise is well known , and discussed e.g. in ref .
the geometry of the problem is that of a half - plane , where a reservoir of the displacing fluid is located at the @xmath19-axis .
the boundary condition of the chemical potential at the half - plane boundary can be connected to the physical effect that is driving the capillary rise . in this paper
we will consider spontaneous imbibition , where the rise is driven by a chemical potential difference in the medium , which favors the displacing fluid @xcite .
this means that the two minima of @xmath10 are at different heights . in our notation
the chemical potential difference is @xmath20 , and we consider chemically homogenous medium , where @xmath21 .
spontaneous imbibition corresponds to a dirichlet boundary condition at the reservoir @xcite .
forced flow , where flow is caused by an imposed mass flux into the system from the reservoir , can be modeled with the neumann boundary condition , where @xmath22 is the flux @xcite .
an analysis along the lines presented in this paper can also be conducted for the case of forced imbibition .
a recent review of phase field modeling of imbibition is given in refs .
@xcite . using the green s function @xmath23 for the 2d laplacian
, equation can be inverted using gauss s theorem .
this leads to the integro - differential form @xmath24 notation here has been shortened by omitting the function arguments , and using unprimed and primed functions for functions of unprimed and primed coordinates , respectively .
the green s functions always take both primed and unprimed coordinates as argument .
also the coordinate invariant form is used , with integration measure given by @xmath25 .
the boundary term @xmath26 vanishes in the case of spontaneous imbibition , or dirichlet boundary condition in half - plane geometry .
using the standard 1d kink solution method for projection to sharp interface @xcite in normal coordinates gives eq .
as @xmath27\\ + \int du\partial_u\phi_0\int ds'du'\sqrt{\det(g')}g\nabla'^2\xi'(\kappa'\partial_{u'}\phi_0'+\alpha ) , \end{split}\ ] ] where the normal coordinates @xmath28 are distances along and perpendicular to the interface , respectively , @xmath29 is the local curvature of the interface , @xmath30 is the surface tension of the phase field model , and finally @xmath31 is the 1d kink solution @xmath32 . in the ginzburg - landau form of @xmath10
this would be given by @xmath33 , with the appropriate choice of dimensionless units .
we have assumed a disorder correlation length that is larger than the interface width , which leads to the constant surface tension obtained . with two further approximations
is approximated with its form for the 1d disorder free system : @xmath34 , where @xmath35 is the normal velocity .
this approximation is similar to the 1d kink solution in the standard procedure , but its range of extent is not only near the interface , but between the reservoir and interface .
another assumption we made in the derivation is neglecting boundary terms of form @xmath36 .
this is justified when the width of the kernel @xmath37 along @xmath19-direction , which is of order @xmath38 , is much larger than the correlation length of the disorder @xmath18 . ]
the standard procedure @xcite can be followed . transforming the equation to cartesian coordinates
is made somewhat more tedious by the necessity to transform derivatives w.r.t . @xmath39 and
@xmath40 , but standard differential geometry methods can be applied . after the sharp interface limit , i.e. @xmath41 , the transformation to cartesian coordinates , and linearization in small fluctuations of the interface @xmath42 and the noise @xmath18 , which also eliminates cross terms proportional to @xmath43 , we get the lie as @xmath44\partial_t\big[h_0(t)+h(x',t)\big]=\\ -\sigma\partial_x^2h(x , t)-\alpha+\frac{\partial_th_0(t)}{m}\xi(x , h_0(t ) ) , \end{split}\ ] ] where the disorder term @xmath45 is given by @xmath46 note that the linearization has been carried out in full here .
this means that the disorder term does not include any dependence on the interface fluctuations .
this eliminates the non - linearity of the quenched noise , which is one of its characteristic properties , but we believe it is not crucial in the regime where the linearization is appropriate . in other words ,
our results show non - trivial features that arise in the effective noise at the interface level with this type of multiplicative bulk disorder , even in the linear regime of weak disorder . the green s function for the dirichlet boundary condition in half - plane geometry is explicitly given by @xmath47 using this , the fourier space representation of the interface equation becomes @xmath48 where @xmath49 , @xmath50 is the fourier transform of the chemical potential difference ( @xmath51 , if @xmath52 , when @xmath53 ) , and the disorder in fourier space is given by @xmath54 in the case of columnar disorder , which does nt depend on @xmath55 , the interface equation simplifies to @xmath56 it is noteworthy that the limit @xmath57 the interface equation is @xmath58 , readily giving the correct washburn law @xcite , if we associate @xmath59 , and @xmath60 .
our method of analysis can be applied to the case of forced flow by simply changing the boundary condition of the phase field model at the reservoir , and applying the corresponding green s function @xcite .
the dispersion relation above is the main result in this section .
it involves two length scales : a crossover length scale @xmath61 @xcite , and the distance from the reservoir @xmath62 .
the deterministic part of the dispersion relation is plotted in fig . [
fig : dispersion ] , at the two limits of these length scales : the limit @xmath63 brings out the `` deep '' limit , @xmath64 , behavior .
the limit @xmath65 shows the `` shallow '' limit , @xmath66 , behavior . a plot from the intermediate regime with @xmath67
is also shown .
( vertical dashed line ) and @xmath62 ( vertical dotted line ) .
the upper figure focuses on the `` deep '' regime , with @xmath63 , in the middle figure these length scales are the same , and the lowest figure focuses on the `` shallow '' regime , with @xmath65.,width=302,height=377 ] the deterministic part of the dispersion relation here is identical to that previously obtained by dub _
@xcite for the case of disorder in the chemical potential . in the `` deep '' limit where @xmath68
, our result also reproduces that of hernandez - machado _ et al . _
@xcite for the one - sided mobility case .
using different methods the same result has also been obtained for the hele - shaw setup by paune and casademunt @xcite , and ganesan and brenner @xcite .
our disorder term in the lie is similar to those obtained in refs .
@xcite in the sense that in all cases the effective noise is linearly proportional to the velocity of the interface propagation . however , the quantitative forms of the noise terms are different when using different methods .
how these differences influence the kinetic roughening of interfaces would need to be determined by extensive numerical comparison between the different results , which at this point has not been conducted . as a linear @xmath69 proportionality in the fourier space representation of the effective noise term
is linked to the @xmath55-derivative of the green s function of the laplacian in real space representation , it appears to us that the linear @xmath69 is present in the noise terms of refs . @xcite and @xcite , but not in ref .
the linear @xmath69 dependence is in general characteristic of effective interface noise caused by conserved bulk disorder @xcite .
however , the @xmath70 dependence ( @xmath71 ) dimensionally cancels the integral over the kernel in the effective noise , eq .. this is explicit in the case of columnar disorder in eq . , but is equally valid with the noise in the non - columnar case .
thus the multiplicative bulk disorder in the mobility leads to a different type of effective noise than the chemical potential disorder , which is considered in @xcite .
dimensionally this can be seen from the definition of the model , eq .
, where the noise term is in front of the gradient of the chemical potential .
the fact that the columnar disorder leads to effective noise , which is local in fourier space , is in accordance with the conclusions of experiments of soriano _ et al . _
@xcite , and with the numerical results from the one - sided model @xcite .
this would indicate that the phase dependence of the mobility is not crucially important when considering the invasion of a viscous fluid into a fluid with negligible viscosity , and when the interface is consequently always stable .
when the direction of the invasion is reversed , as studied recently with the one - sided model in @xcite , the situation is naturally quite different .
while the stochastic mobility case of the previous section is heuristically appealing , a more faithful treatment of the wall disorder should start from the microscopic roughness of the walls . to this end , in this chapter we will study a 3d version of the same phase - field model as the 2d model , but where the mobility is constant and the disorder is explicitly included as fluctuations of the wall position .
thus the geometry of the model is that of a hele - shaw cell : the 3d volume between two walls that are planar on average , but fluctuate .
we will show here that by proper mathematical formulation this model can indeed be analyzed by a generalized projection formalism @xcite .
the basic idea is to perform a mathematical transformation from the basic cartesian to a local curvilinear coordinate system as defined by the wall itself . to this end , we consider the one - wall setup as shown in fig .
2 . the one - wall setup neglects the meniscus - mediated interaction between the two contact lines at the two walls .
the one - wall approximation also neglects finite gap spacing , i.e. the distance between the two walls in the hele - shaw cell , which fluctuates as result of the wall fluctuations .
this induces additional disorder effects when the wall fluctuations are comparabale to the gap spacing , but it remains to be studied if the gap effect can be separated from the contact line interaction , which would be represented by two coupled equations of motion for the two contact lines . in the present work , we only consider to the one - wall approximation , or the limit of large gap spacing .
disorder at the wall surface is taken into account by describing local corrugations in the wall position around @xmath72 by a ( small ) function @xmath73 .
the explicit coordinate transformation to the local , curvilinear wall coordinate system is defined by @xmath74 which corresponds to @xmath75 when @xmath73 .
this means that in the new coordinate system the wall is located back at @xmath75 .
given the proper green s function , @xmath76 the phase field equation can be inverted in any geometry and coordinate system as @xmath77 where @xmath78 is the corresponding surface term , and @xmath79 is the volume element for the coordinate system .
the green s function appropriate for the above mentioned coordinate system is considered in some detail in appendix [ curvy ] . here
we compute the correction to the original cartesian green s function to linear order in @xmath80 .
the final result we obtain , after neglecting some surface contributions that are discussed in more detail in appendix [ curvy ] , is what one would expect by simply plugging the above definitions into the cartesian green s function and linearizing in @xmath81 : @xmath82 here @xmath83 is the green s function for the laplacian in 3d cartesian coordinates as given in eq . .
here we again only consider spontaneous capillary rise , where the boundary conditions for the phase field model are zero chemical potential at the reservoir and zero flux at the walls .
thus the surface integral term in eq .
is identically zero .
the projection and linearization of the integral equation follows the standard projection operation theory @xcite , which we already used in the previous chapter for the 2d model .
the generalization for the present case is straightforward . after projection the integral equation is expressed in terms of the 2d meniscus variable @xmath84 and has the following form : @xmath85 when linearizing the above equation , it must be done simultaneously in the meniscus fluctuations , _ i.e. _ @xmath86 , and in the wall fluctuations using the linearized green s function of eq .. this results in the linearized green s function evaluated at the meniscus : @xmath87 substituting this into the meniscus equation gives @xmath88 where the left hand side equation is to the zeroth order , and the right hand side is to the first order in @xmath89 or @xmath90 .
these terms are defined in the same fashion as those in cartesian coordinate system @xcite .
the terms @xmath91 and @xmath92 arise from the fluctuating wall .
they are given by @xmath93 the zeroth order equation would give the washburn law , if we used the green s function for the geometry between two walls and assumed a constant curvature for the meniscus .
we will assume an average profile @xmath94 , which can be considered to obey washburn s law even though we have only a single vertical wall in the system . since @xmath62 is not needed for determining the form of the evolution equation for the fluctuating part @xmath42 of eq .
at that single wall , the precise time - dependence of @xmath62 is not crucial for the analysis to be presented below .
a local equation of motion for the meniscus fluctuations can be obtained by fourier - cosine transformation following @xcite .
the above terms become @xmath95&= & \frac{1}{2}\dot{h}_0h(\vec{k},t ) ; \\ { \mathcal{f}}_{x / k_x}{\mathcal{f}}^{cos}_{y / k_y}[i_{c}]&= & \frac{1}{2}e^{-2kh_0}\dot{h}_0h(\vec{k},t ) ; \\ { \mathcal{f}}_{x / k_x}{\mathcal{f}}^{cos}_{y / k_y}[i_{d}]&= & \frac{1}{2k}\dot{h}(\vec{k},t)\left(1-e^{-2kh_0}\right ) ; \\
i_{\textrm e}&=&0 ; \\
{ \mathcal{f}}_{x / k_x}{\mathcal{f}}^{cos}_{y / k_y}[i_{\textrm f}]&=&\frac{\dot{h}_0(t)}{2k } \left(1-e^{-2kh_0}\right)\delta h(k_x , h_0(t)),\end{aligned}\ ] ] where @xmath96 .
we then have the meniscus equation of motion using the above in the fourier transform of eq .
: @xmath97 the deterministic part of the above meniscus equation is identical to the deterministic part of the lie derived from the 2d phase - field model , apart from the dimensionality .
this is by construction , since the same method was used for the same theory in different dimensions by applying the corresponding green s functions . a similar analysis
can also be performed for the case where the disorder at the walls consists of chemical impurities ( _ i.e. _ spatially fluctuating surface tension ) instead of spatial roughness @xcite . in this case
the deterministic part of the meniscus equation is by construction identical to that of the above .
however , there is no effective noise at the meniscus level , since the effect of the disorder comes in from the contact line that serves as a boundary condition for the meniscus . to proceed to the level of the 1d contact line we consider the generalized variational approach @xcite .
formally , one can write the 3d phase field model in terms of variations of a rayleigh dissipation functional , and a free energy functional . then , using approximations that express higher dimensional entities in term of the relevant lower dimensional ones , we obtain a chain of projection equations as @xmath98 } { \delta \dot { \phi}(x , y , z;t ) } & = & - \frac { \delta f_{\textrm { 3d}}[\phi ] } { \delta \phi(x , y , z;t ) } \\ \label{2d_functional_eom } \rightarrow \frac { \delta r_{\textrm { 2d } } [ \dot h ] } { \delta \dot h ( x , y;t ) } & = & - \frac { \delta f_{\textrm { 2d}}[h ] } { \delta h(x , y;t ) } \\ \rightarrow \frac { \delta r_{\textrm { 1d } } [ \dot c ] } { \delta \dot c ( x;t ) } & = & - \frac { \delta f_{\textrm { 1d}}[c ] } { \delta c(x;t)}\ , , \label{ray}\end{aligned}\ ] ] where @xmath99 refers to the rayleigh dissipation functional , and @xmath100 refers to the free energy functional in @xmath101 dimensional space . here
the relevant 3d , 2d and 1d objects are the phase field , the meniscus profile and the contact line profile , respectively .
the variable @xmath102 denotes the fluctuating contact line profile , and @xmath103 for the one - wall case .
the corresponding expansion for the contact line is @xmath104 . for small fluctuations @xmath42 and @xmath105
, consistency requires that @xmath106 . the projection from 3d to 2d
is made possible by the 1d kink approximation in the direction normal to the interface , as demonstrated in the preceding section .
the corresponding approximation we have used to make the 2d to 1d projection possible is the quasi - stationary ( qs ) approximation @xmath107 ; @xmath108 , which corresponds to the minimum of energy constrained by the contact line profile .
the meniscus can then be expressed in terms of the contact line as @xmath109 the explicit forms for the rayleigh dissipation and free energy functionals that reproduce the meniscus equation when plugged into eq .
are @xmath110&=&\int_{-\infty}^{\infty}~dx_1dx_2\int_0^{\infty}~dy_1dy_2\int~dt_1 \dot{h}(x_1,y_1,t_1);\nonumber\\ & & \times \tilde{g}_{3d}(x_1,y_1,h(x_1,y_1,t_1);x_2,y_2,h(x_2,y_2,t))\dot{h}(x_2,y_2,t_1 ) \label{2d_rayleigh}\\\hspace{2 cm } f_{2d}[h]&=&\sigma_b\int_{-\infty}^{\infty}~dx_1\int_0^{\infty}~dy_1\int~dt_1~\sqrt{1+|\nabla h(x_1,y_1,t_1)|^2}.\label{2d_energy}\end{aligned}\ ] ] the effective 1d functionals can be obtained from the above by inserting the quasi - stationary approximation @xmath111 into and . in order to obtain
the 1d equation of motion to linear order in small fluctuations one needs to expand the functionals to second order in both @xmath112 and @xmath113 , and then take the variation with respect to the contact line as shown in eq .. neglecting the zeroth order equation for the reasons mentioned earlier ,
the general fourier space equation of motion we obtain for the first order fluctuations is @xmath114=-\sigma_b |k_x| c(k_x , t),\ ] ] where the lhs is the variation of the rayleigh dissipation functional , and the rhs is the variation of the free energy .
the rhs is recognized as the deterministic restoring force acting on the contact line @xcite .
the shorthand notations stand for @xmath115 not all of these integrals are solvable in closed form , but can be approximated to a good degree of accuracy by the following expressions : @xmath116&=&\frac{\dot{c}_0}{2|k_x|}c(k_x , t ) ; \label{eq_3d_corrug_first_cleom_term}\\ { \mathcal{f}}_{x / k_x}[i_{3}]&=&\frac{2\dot{c}_0}{\pi \frac{e^{-2|k_x|c_0s}}{s^3\sqrt{s^2 - 1}}\nonumber\\ & & \approx \frac{1.14\cdot 4}{3\pi { \mathcal{f}}_{x / k_x}[i_{4}]&=&\frac{2\dot{c}(k_x , t)}{k_x^2\pi}\int_{1}^{\infty}ds \frac{1- e^{-2|k_x|c_0s}}{s^4\sqrt{s^2 - 1}}\nonumber\\ & & \approx \frac{4\dot{c}(k_x , t)}{3\pi k_x^2 } \left(1-e^{-2.28|k_x|c_0}\right);\\ { \mathcal{f}}_{x / k_x}[i_{5}]&=&0;\\ { \mathcal{f}}_{x / k_x}[i_{6}]&=&\frac{2}{\pi |k_x|}\dot{c}_0\delta h(k_x , c_0 ) \int_1^\infty ds \frac{1- e^{-2|k_x|c_0s}}{s^2\sqrt{s^2 - 1}}\nonumber\\ & & \approx\frac{2}{\pi |k_x|}\dot{c}_0\delta h(k_x ,
c_0 ) \left(1- e^{-2.56|k_x|c_0}\right).\end{aligned}\ ] ] we note that the corrections to the free energy functional in the curvilinear coordinates are of third order in @xmath81 and @xmath42 .
this can be seen by coordinate transforming the area element , which in cartesian coordinates is @xmath117 .
finally , after approximating @xmath118 and @xmath119 , the equation of motion for the contact line fluctuations becomes @xmath120 note that all of the approximations above are for dimensionless quantities , with errors depending on the physical parameter @xmath121 .
the relative errors in the approximated functional forms are under 3% , when compared against numerical integration of the respective integrals for different values of @xmath122 .
an exception to this are relative errors of @xmath123 $ ] and @xmath124 $ ] when @xmath125 , as both @xmath126 and @xmath127 vanish at the limit , causing the relative error behave badly .
however , at machine precision away from @xmath128 then these errors are no more than 15% , and more importantly the error of the complete dispersion relation stays within the 3% error margin .
this is due to the fact that the dispersion remains finite , as one can see from figure [ fig : dispersion ] .
apart from simple numerical factors , the contact line equation above has the same functional form as the results derived in the previous sections . in particular , @xmath129 in the `` deep '' limit @xmath130 , which thus agrees with the previous works discussed earlier @xcite .
this form of dispersion relation is always obtained by our method for interfaces in model b. this has to do with the quasi - stationary approximation , which essentially assumes that meniscus fluctuations dampen quickly in the direction perpendicular to the contact line , in order to obtain temporally local equations .
how this leads to the coupling of the meniscus and contact line dynamics is discussed in more detail in another publication @xcite .
the effective noise term we obtain from the 3d model shares the property of linear dependence on the velocity of the propagation with the 2d mobility noise , and with the previous analyses @xcite . in the case of surface tension impurities at the wall @xcite the effective contact line noise is proportional to @xmath131 , whereas in the fluctuating wall case in eq .
the @xmath132 dependence is linear .
this is analogous to the 2d mobility disorder in the sense that the effective noise is different from that obtained for conserved disorder .
the extra factor of @xmath133 as compared to the 2d model ( eq . ) comes from the fact that the disorder in the 3d model comes from the walls , whereas in the 2d model the disorder is in the bulk .
we note that the more complicated properties of the noise in the form of non - locality in fourier space are lost by our approximations .
note that the noise is still non - local in real space , as is apparent from its real space representation @xmath134 in eq . .
in addition to fourier space non - localities , our one - wall approach does nt explicitly include the gap spacing , which provides an additional physical length scale . in spite of this , our results are in accordance with those of ref .
@xcite , where the gap fluctuations were considered as the only source of disorder in context of darcy s law .
in this paper , we have studied the effective interface dynamics of the three - phase contact line in a hele - shaw experiment by deriving the meniscus and contact line equations of motion from higher dimensional bulk phase - field theories by projection methods .
the projection methods take into account the non - local dynamics of the system caused by local mass conservation , and can be systematically applied starting from a full 3d description .
we have here considered two particular models , namely an _
ad hoc _ model where the disorder is in the effective mobility in 2d @xcite , and a more microscopic model where wall corrugation in 3d is explicitly treated with a curvilinear coordinate transformation . in both cases , we have focused on the limit of small disorder by linearizing in disorder strength and in the fluctuations caused by the disorder . by construction
this linearization , performed in real space , causes the fourier space representations of the equations of motion to be local .
the upside of this is that the effective dynamics are written in a concise manner , and the physical predictions are easily interpreted and the equations we obtain are readily affable to numerical analysis .
the obvious downside is that the procedure involves a number of approximations , the validity of which is not certain _ a priori_. in particular , the quasi - stationary approximation of eq .
, which ultimately enables our contact line analysis , requires a critical assessment .
a more rigorous approach would in fact consider the contact line as the boundary condition to the meniscus equation of motion .
however , explicitly solving the meniscus profile as a function of the contact line leads to an equation that is , among other things , non - local in time .
thus we are forced to simplify the model by using the qs approximation , the validity of which we can consider both from a physical perspective , or more rigorously by considering the limits of the meniscus equation of motion .
physically , the qs approximation comes from the minimum of meniscus energy constrained by the boundary condition of the contact line .
this is expected to define the meniscus profile when the meniscus moves slowly , and thus it s called the quasi stationary approximation .
mathematically the meniscus equation reduces to the diffusion equation when @xmath135 , and @xmath136 . in this limit
the meniscus level disorder @xmath137 acts as a source term @xmath138 this leads to an additional disorder term in the qs approximation , which then leads to a plethora of new first order disorder terms in @xmath139 and @xmath140 .
however , all these new disorder terms arising from @xmath139 are proportional to @xmath141 , and thus not relevant in the qs limit .
additionally , the two new disorder terms created in @xmath140 due to @xmath137 cancel each other out exactly .
thus , we expect our results with the simplified version of the qs approximation to hold in this particular limit .
in addition to the detailed derivations and new projection formalism presented here , our purpose has been to quantitatively compare two different approaches to modeling rough wall hele - shaw experiments , namely that based on the 2d phase field model with a stochastic mobility , eq . , and the 3d phase field model with a fluctuating geometry .
the projection method we use for both cases produces the linear response of the meniscus and contact line to small fluctuations . for both cases ,
the @xmath142 dependence of the meniscus and contact line deterministic lies is the same .
in particular , in the special case of the `` deep '' limit where @xmath143 , the asymptotic forms of our general dispersion relations are in agreement with previous works on the hele - shaw problem by paune and casademunt @xcite , ganesan and brenner @xcite and hernandez - machado _ et al .
the main advantage of our method is the way it incorporates the noise into the projection , and thus allows us to study the effective noise caused to the contact line level by bulk or wall disorder .
the main result of this analysis is that in both cases the effective noise is linearly proportional to the velocity of the interface .
while this result qualitatively agrees with the other works cited above , quantitative differences remain in the form of the noise terms .
the relevance of these differences to the actual kinetic roughening of the interfaces remains a challenging numerical problem .
we would like to thank k. elder for inspirational and pleasant discussions , as well as for sharing his expertise and results on the sharp interface analysis .
this work has been supported in part by the academy of finland through its center of excellence grant ( comp ) .
s.m . has been supported by a personal grant from the academy of finland .
in this appendix we will consider in more detail the green s function of the laplacian in the fluctuating coordinate system @xmath144 which is schematically depicted in figure [ schem ] .
the generalization to the two - wall setup is straightforward , but very tedious including two independent disorder functions . in particular
, we will consider the correction to the green s function to linear order in small wall fluctuations @xmath80 .
first , the metric tensor of the above coordinate system can be obtained by transformation of the cartesian metric tensor as @xmath145=\sum_i \frac{\partial x^{i}}{\partial x^{i ' } } \frac{\partial x^{i}}{\partial x^{j'}}= \begin{bmatrix } 1+(\partial_x\delta h)^2 & \partial_x\delta h & \partial_z\delta h\partial_x\delta h\\ \partial_x\delta h & 1 & \partial_z\delta h\\ \partial_z\delta h\partial_x\delta h & \partial_z\delta h & 1+(\partial_z\delta h)^2 \end{bmatrix}.\ ] ] the coordinate transformation from cartesian coordinates is merely a shift , and thus the integration measure should not change , meaning that the jacobian in the coordinate transformation of an integral should be unity .
this is indeed so , since @xmath146)\equiv 1.\ ] ] in the case of cartesian coordinates we can obtain the green s function , which we denote @xmath147 , using the image charge method with the dirichlet boundary condition at @xmath148 and the neumann boundary condition at @xmath149 : @xmath150 @xmath151.\ ] ] we work in the limit of small fluctuations , so we write the laplacian in the curvilinear coordinates as the cartesian laplacian plus a correction , @xmath152 .
note that we unconventionally denote @xmath153 for any coordinates @xmath154 $ ] .
the correction @xmath155 is explicitly shown below : @xmath156
\frac { \partial^2}{\partial x \partial y } + 2 \left [ \frac { \partial \delta h(x , z ) } { \partial z } \right ] \frac { \partial^2}{\partial z \partial y } + [ \frac { \partial^2 \delta h(x , z ) } { \partial x^2 } + \frac { \partial^2 \delta h(x , z ) } { \partial z^2 } ] \frac { \partial } { \partial y}. \label{correctionoperator}\end{aligned}\ ] ] in order to use eq .
for the curvilinear coordinates , we need the green s function , which has the property of @xmath157 \tilde { g}_{\textrm { 3d } } ( \boldsymbol { r}',\boldsymbol { r_1 } ' ) = - \delta ( \boldsymbol { r } ' - \boldsymbol { r_1}')$ ] .
since the laplacian in the curvilinear coordinates can be expressed as the cartesian laplacian plus a correction , we can find the inverse of the curvilinear laplacian , or @xmath158 , to first order in @xmath81 as @xmath159 the above operator notation can be written in full form as : @xmath160 substituting @xmath147 into the above , we can work out the correction to the green s function . after using a similar argument to neglect surface integrals of type @xmath36 as we did with eq .
, we find @xmath161 at this point the primes can just be dropped , since @xmath162 .
this result is hardly surprising , since a simple substitution of @xmath163 to @xmath164 yields identical results to linear order .
the neglected surface integrals include a reservoir term and a wall term .
the reservoir term can be readily seen to be small when the meniscus is further away from the reservoir than the disorder correlation length .
additionally , the reservoir boundary correction is zero when considering columnar _
i.e. _ @xmath165-independent disorder .
the wall term is more problematic , since it involves the boundary correction due to fluctuation in the direction of the wall normal .
we have to observe the meniscus further away from the wall than the disorder correlation length in order to neglect this boundary correction .
the absence of boundary disorder corrections is highly desirable if we are to keep our formalism tractable , so we have neglected the boundary corrections to the green s function . | we consider the influence of quenched noise upon interface dynamics in 2d and 3d capillary rise with rough walls by using phase - field approach , where the local conservation of mass in the bulk is explicitly included . in the 2d case the disorder is assumed to be in the effective mobility coefficient , while in the 3d case we explicitly consider the influence of locally fluctuating geometry along a solid wall using a generalized curvilinear coordinate transformation . to obtain the equations of motion for meniscus and contact lines ,
we develop a systematic projection formalism which allows inclusion of disorder . using this formalism
, we derive linearized equations of motion for the meniscus and contact line variables , which become local in the fourier space representation .
these dispersion relations contain effective noise that is linearly proportional to the velocity .
the deterministic parts of our dispersion relations agree with results obtained from other similar studies in the proper limits . however , the forms of the noise terms derived here are quantitatively different from the other studies . |
the interest in the study of quantum many - body systems from the modern standpoint of quantum field theories @xcite has grown since the 1960 s , ranging from low to extremely high energy and temperature regimes @xcite . in 1975 ,
takahashi and umezawa proposed an operator - based formalism called thermofield dynamics ( tfd ) @xcite in order to describe quantum systems at finite temperature in thermal equilibrium . in the same year
the discovery of the hawking effect @xcite , from the black hole thermal radiation , was one of the most innovative results in quantum field theory in curved space - time .
later on , davis @xcite suggested that a similar effect would occur even in a flat space - time . in 1976 , unruh @xcite showed that the essential characteristics of the hawking effect were contained in a simpler situation : an accelerated detector , would be excited by particles in vacuum , property known as the unruh effect .
this discovery of the hawking - unruh effect has inspired many researchers , generating since them a series of important results , @xcite .
there are two traditional approaches to the unruh effect , one is through the quantization of non - massive field in curvilinear coordinates , the other , due to dewitt @xcite and which we shall use in this work , is considering the acceleration as a hyperbolic parametrization in minkowski space time .
the present work aims at investigating the manifestation of the unruh effect , approaching the issue from yet another perspective @xcite .
we consider a two - level system in interaction with a non - massive scalar field , for which we deduce a master equation from von neumann equation .
subsequently , we present the thermal states of thermofield dynamics ( tfd ) in the number representation and use it to thermalize the theory via vacuum expected values , singling out in a clearer way the unruh temperature . the original proposal of unruh and dewitt , considers a detector as a point object that is linearly coupled with a zero mass scalar field through an interaction of monopole type @xmath0 where @xmath1 is a coupling constant and @xmath2 is the monopole moment of the detector .
specifically , @xmath3 is the hamiltonian of the detector which has a discrete spectrum @xmath4 we assume that the detector moves along the world line described by the coordinate @xmath5 , where @xmath6 is the proper time of the detector .
assume that at an initial instant @xmath7 the detector is in the state @xmath8 and the field in the vacuum @xmath9 .
for an arbitrary trajectory , it is reasonable to assume that the detector will not remain in the initial state , but undergoes a transition to an excited state .
therefore , for a later time @xmath10 , the detector is in the state @xmath11 and field in the state @xmath12 . for @xmath1
sufficiently small , we can calculate via first order perturbation theory the transition probability @xmath13 the wightman function in the integrand is well known and can be explicitly calculated , giving @xmath14}.\label{eq : funcao.wightman}\ ] ] firstly , we consider the case in which the detector follows a inertial world line , i.e. , the parametric equations @xmath15 from which we obtain for the wightman function , @xmath16 and so the probability @xmath17 the integral in turns out to be @xmath18 . therefore , the transition probability to an excited state is zero , since , as expected , a detector at rest must not spontaneously excite .
now , consider the case where the detector is uniformly accelerated ; this is equivalent to a hyperbolic world line , parametrized according to @xmath19 similarly , for the wightman function , @xmath20^{2}}\mbox{,}\label{eq : wightman.somatorio}\ ] ] so that the resulting probability is @xmath21^{2}}.\label{eq : prob.transi=0000e7ao.acelerado}\ ] ] after integration , it turns out to be @xmath22 from which one recognizes the factor @xmath23^{-1}$ ] as being the planck distribution associated with the unruh temperature @xmath24 where @xmath25 is the boltzmann constant .
in this section we briefly present the formalism introduced by takahashi and umezawa @xcite for a real time thermal field theory , where statistical averages of physical observables correspond to the expected values in a thermal bogoliubov vacuum . in schwinger s measurement algebra ( sma )
an operator is defined as @xmath26 where @xmath27 is known as the measurement symbol or projector into the @xmath28 basis ; defining the composition rule @xmath29 in such a context , the expectation value of a given observable @xmath30 in the single @xmath31 basis is the scalar @xmath32\mbox{.}\ ] ] from that one , we can define the statistical average @xmath33=\underset{m}{\sum}\pi\left(m\right)\left\langle a\right\rangle _ { m}\mbox{,}\ ] ] where @xmath34 are the statistical weights and we can recognized the density operator in the sma as @xmath35 it is known that the inclusion of temperature may cause the duplication of the system degrees of freedom @xcite .
we can achieve this by the introduction of the relation @xmath36 once this is done , we can rewrite the density operator as @xmath37\left[\underset{m}{\sum}\sqrt{\pi\left(m\right)}m\left(\widetilde{m},m\right)\right]\equiv\left|0_{\left(\beta\right)}\right\rangle \left\langle 0_{\left(\beta\right)}\right|\mbox { , } \end{array}\ ] ] where we have made use of the composition law of composite measurement symbols . in thermal equilibrium , the statistical weights assume the form @xmath38 . in the above expression , we can interpret the density operator as a projector into a thermal vacuum state @xmath39 . for a quantum system in thermal equilibrium at temperature @xmath40 , in the canonical ensemble , the statistical average of an observable a is then given by @xmath41=\left\langle 0_{\left(\beta\right)}\left|a\right|0_{\left(\beta\right)}\right\rangle \mbox{,}\label{eq : media.estatistica}\ ] ] where @xmath42 @xmath43 in order to construct this representation it s appropriated to introduce a fictitious system identical to the original system , so that @xmath44 belongs to this new system .
for the free scalar field , we have @xmath45^{\frac{1}{2}}\underset{\mathbf{k}=0}{\overset{\infty}{\sum}}e^{-\frac{1}{2}\beta\omega_{\mathbf{k}}n_{\mathbf{k}}}\left|n_{\mathbf{k}},\widetilde{n}_{\mathbf{k}}\right\rangle \mbox{,}\ ] ] from which follows the bose - einstein distribution @xmath46^{-1}\mbox{.}\label{eq : est.bos.einstein}\ ] ]
in this section we outline the steps leading to the finite temperature master equation following closely the procedure presented in @xcite ( see also @xcite ) . consider a microscopic system @xmath30 surrounded by a dissipative environment @xmath47 .
the typical hamiltonian that describes the total system , in the schrdinger representation , is @xmath48 where @xmath49 is the free hamiltonian for the system @xmath30 , @xmath50 corresponds to that for the _ reservoir _ @xmath47 and @xmath51 is the interacting potential . using the density operator formalism to describe
the total system , the dynamical evolution , in the interaction representation , will be given by the liouville - von neumann equation , @xmath52\mbox{.}\label{eq : lvn}\ ] ] iterating the liouville - von neumann equation to second order , gives @xmath53+\left(\frac{1}{i\hbar}\right)^{2}\int_{t}^{t+\delta t}dt'\int_{t}^{t'}dt''\left[v\left(t'\right),\left[v\left(t''\right),\rho\left(t''\right)\right]\right]\mbox{.}\ ] ] assuming that the interacting potential is sufficiently weak , the perturbation caused in the _ reservoir _ by the microscopic system is negligible and we can consider that the _ reservoir _ is in the stationary state @xmath54 where @xmath55 is the reduced density operator for the _
reservoir_. thus , the density operator for the total system assumes the form @xcite
@xmath56 the resulting master equation for the reduced density operator in the markov - born approximation , or equivalently , the coarse - graining approximation for its time evolution is @xmath57\right]\mbox{.}\ ] ] assuming a separable potential @xmath58 and expanding the master equation commutator we can identify the function @xmath59\mbox{.}\label{eq : g(t)}\ ] ] this is the two - point correlation function and we will use it to introduce temperature in the theory by making the substitution @xmath60 so that becomes @xmath61 finally , we can derive the finite temperature master equation in the coarse - graining approximation , @xmath62 we consider the less common case of a fermi oscillator and how its population , energy levels , @xmath63 behaves ; let the system @xmath30 be a fermion oscillator of frequency @xmath64 interacting with a _ reservoir _ @xmath47 , composed by @xmath28 independent oscillators of frequency @xmath65 in thermal equilibrium .
so , we define the system total hamiltonian by @xmath66 where the hamiltonian
@xmath67 belongs to the hilbert space @xmath68 , and @xmath69 where the respective ladder operators obey the usual _ anticommuting _ algebra , @xmath70 in order to describe the interaction with the _ reservoir _ , we introduce the potential @xmath71\mbox{,}\ ] ] where @xmath72 is a coupling parameter and @xmath73 , @xmath74 non - negative constants , the expression above being the most general sesquilinear interaction . making @xmath75 and assuming the rotating wave approximation , @xmath76
, we arrive at @xmath77 where @xmath78 writing the master equation commutator explicitly , we have @xmath79\right]\\ \\
= \left(b\left(t'\right)b^{\dagger}\left(t''\right)\sigma_{a}\left(t\right)-b^{\dagger}\left(t''\right)\sigma_{a}\left(t\right)b\left(t'\right)\right)tr\left[\sigma_{r}r^{\dagger}\left(t'\right)r\left(t''\right)\right]\\ \\ + \left(\sigma_{a}\left(t\right)b\left(t''\right)b^{\dagger}\left(t'\right)-b^{\dagger}\left(t'\right)\sigma_{a}\left(t\right)b\left(t''\right)\right)tr\left[\sigma_{r}r^{\dagger}\left(t''\right)r\left(t'\right)\right]\\ \\ + \left(b^{\dagger}\left(t'\right)b\left(t''\right)\sigma_{a}\left(t\right)-b\left(t''\right)\sigma_{a}\left(t\right)b^{\dagger}\left(t'\right)\right)tr\left[\sigma_{r}r\left(t'\right)r^{\dagger}\left(t''\right)\right]\\ \\
+ \left(\sigma_{a}\left(t\right)b^{\dagger}\left(t''\right)b\left(t'\right)-b\left(t'\right)\sigma_{a}\left(t\right)b^{\dagger}\left(t''\right)\right)tr\left[\sigma_{r}r\left(t''\right)r^{\dagger}\left(t'\right)\right]\mbox{. } \end{array}\ ] ] by ensuring that the vacuum of the system is separable @xmath80 we also guarantee the separability of the function @xmath81 , @xmath82=\left\langle 0_{\left(\beta\right)}\right|r\left(t'\right)r\left(t''\right)\left|0_{\left(\beta\right)}\right\rangle \\ \\
= \left\langle 0_{\left(\beta\right)}\right|r\left(t'\right)r\left(t''\right)\left|0_{\left(\beta\right)}\right\rangle _ { 1}\left\langle 0_{\left(\beta\right)}\right|r\left(t'\right)r\left(t''\right)\left|0_{\left(\beta\right)}\right\rangle _ { 2}\cdots\mbox{. } \end{array}\ ] ] the fermion traces follow from the above expression , in which there appears , as expected , the fermi - dirac distribution , @xmath83=\underset{i}{\sum}\left|g_{i}\right|^{2}e^{-i\omega_{i}\left(t''-t'\right)}\left(e^{\beta\hbar\omega}+1\right)^{-1}\mbox{,}\\ \\ tr\left[\sigma_{r}r^{\dagger}\left(t''\right)r\left(t'\right)\right]=\underset{i}{\sum}\left|g_{i}\right|^{2}e^{-i\omega_{i}\left(t'-t''\right)}\left(e^{\beta\hbar\omega}+1\right)^{-1}\mbox{,}\\ \\ tr\left[\sigma_{r}r\left(t'\right)r^{\dagger}\left(t''\right)\right]=\underset{i}{\sum}\left|g_{i}\right|^{2}e^{-i\omega_{i}\left(t'-t''\right)}\left[1-\left(e^{\beta\hbar\omega}+1\right)^{-1}\right]\mbox{,}\\ \\
tr\left[\sigma_{r}r\left(t''\right)r^{\dagger}\left(t'\right)\right]=\underset{i}{\sum}\left|g_{i}\right|^{2}e^{-i\omega_{i}\left(t''-t'\right)}\left[1-\left(e^{\beta\hbar\omega}+1\right)^{-1}\right]\mbox{. } \end{array}\ ] ] under these considerations , we arrive at the equation for the evolution of the population of a given state @xmath28 , for the microscopic system , @xmath84 where @xmath85\mbox{,}\\ \\ t_{f } & \equiv2\left(\frac{1}{\hbar}\right)^{2}\underset{i}{\sum}\left|g_{i}\right|^{2}\omega_{0i}^{-2}\left[\frac{1-cos\left(\omega_{0i}\delta t\right)}{\delta t}\right]\left(e^{\beta\hbar\omega_{i}}+1\right)^{-1}\mbox{. } \end{array}\ ] ] interpreting the above equation , we realize that @xmath86 is associated to the spontaneous emission and @xmath87 to the process of stimulated absorption and emission .
more precisely , @xmath88 is the spontaneous emission rate between the states @xmath89 and@xmath90 , so the state @xmath89 decays at a rate @xmath88 while the state @xmath90 is populated at rate @xmath91 . similarly , there are processes of absorption and stimulated emission , at rates @xmath92 and @xmath93 , respectively . observing the thermodynamic limit of transition rates for low temperatures , @xmath94
we find that the only allowed variations are the fluctuations .
however , in the high temperatures limit , @xmath95 the transition rate takes the value @xmath96 , due to restrictions imposed by the fermion algebra , allowing the excitation of the only mode of energy available . at progressively higher temperatures , the fermion particles of the thermal bath exchange energy with the microscopic particle through short - range correlations , exciting it or inducing it to decay to their only two modes of oscillation , with equal probabilities . finally , if we want to describe a boson oscillator in a fermion heat bath we use the expression without restricting the accessible states . in the fermion case
having only two allowed modes we can easily see how the energy for the microscopic system evolves @xmath97=tr\left[\left(\frac{d}{dt}\sigma_{a}\left(t\right)\right)h_{a}\right]\\ \\ = \underset{n}{\sum}\left\langle n\left|\left(\frac{d}{dt}\sigma_{a}\left(t\right)\right)h_{a}\right|n\right\rangle = \underset{n}{\sum}\hbar\omega_{0}\left(n-\frac{1}{2}\right)\left\langle n\left|\left(\frac{d}{dt}\sigma_{a}\left(t\right)\right)\right|n\right\rangle \\ \\
= -\hbar\omega_{0}\frac{1}{2}\left\langle 0\left|\left(\frac{d}{dt}\sigma_{a}\left(t\right)\right)\right|0\right\rangle + \hbar\omega_{0}\frac{1}{2}\left\langle 1\left|\left(\frac{d}{dt}\sigma_{a}\left(t\right)\right)\right|1\right\rangle \\ \\ = \hbar\omega_{0}\left(-c\sigma_{a_{1,1}}\left(t\right)+t_{f}\sigma_{a_{00}}\left(t\right)\right)\mbox{. } \end{array}\ ] ] if we look for a stationary state @xmath98 we find that we can only reach such state with @xmath99 , so the microscopic system energy stays oscillating in a local non equilibrium situation ; this is fundamental and enables us to construct the master equation . in order to go further and integrate the equation ,
it is necessary to know the form of @xmath100 .
for this reason , we propose a two - level atomic system described by the hamiltonian @xmath101 where @xmath102 and @xmath6 is the proper time of the atom , whose eigenvalues are @xmath103 the proposed field - atom interaction is similar to and given by the hamiltonian @xmath104 where @xmath1 is the coupling constant and @xmath105 , with @xmath106 and @xmath107 .
it is noteworthy that the operators @xmath108 and @xmath109 can be seen as raising and lowering operators of the atomic system and , together with @xmath110 , generate the @xmath111 algebra , @xmath112=2r_{3},\label{eq : rela=0000e7ao.comuta=0000e7ao.r+.r-}\ ] ] @xmath113=\pm r_{\pm}\mbox{.}\label{eq : rela=0000e7ao.comuta=0000e7ao.r3.r+-}\ ] ] from the heisenberg equation @xmath114\mbox{,}\label{eq : eq.heisenberg}\ ] ] where @xmath115 , we obtain the evolution equations @xmath116\mbox{,}\label{eq : eq.mov.r+}\ ] ] @xmath117\mbox{.}\label{eq : eq.mov.r3}\ ] ] working on and to first order in @xmath1 we can conveniently breaking then down into a free part , which preserves in the absence of coupling , @xmath118 @xmath119 and another part of the resulting coupling @xmath120\mbox{,}\label{eq : r+.source}\ ] ] @xmath121\mbox{,}\label{eq : r3.source}\ ] ] where @xmath122 is the free part of the field , @xmath123 , @xmath124\mbox{,}\label{eq : eq.campo.part.livre}\ ] ] @xmath125\mbox{.}\label{eq : eq.campo.part.intera=0000e7ao}\ ] ] therefore , we write the atom energy variation , @xmath126 , as @xmath127\mbox{.}\label{eq : evolu=0000e7ao.h.a}\ ] ] one of the purpose of this section is to identify and quantitatively analyze the contributions from the vacuum fluctuations and radiation reaction , so , we can be identified that the free part of the filed is related to the vacuum fluctuations and , consequently , the radiation reaction is caused by the interaction part .
the energy is a physical observable and , as such , represented by a hermitian operator , we would like to recognize these two different contributions on the energy variation and associates them with two physical processes which are represented by hermitian operators .
for this condition to be valid it is necessary to perform the ordering of operators in .
acknowledging that in quantum theory exists one indeterminacy as to the order of operators , lets consider the product @xmath128 , resulting for the vacuum fluctuation @xmath129\right\ } \\ \\
+ \left(1-\lambda\right)\left\ { -\mu\frac{d}{d\tau}r_{2}^{f}\left(\tau\right)\phi^{f}\left(\tau\right)-i\mu^{2}\intop_{\tau_{0}}^{\tau}d\tau'\phi^{f}\left(\tau'\right)\phi^{f}\left(\tau\right)\frac{d}{d\tau}\left[r_{2}^{f}\left(\tau'\right),r_{2}^{f}\left(\tau\right)\right]\right\ } \mbox { , } \end{array}\ ] ] and for the radiation reaction @xmath130\left[\lambda r_{2}^{f}\left(\tau'\right)\frac{d}{d\tau}r_{2}^{f}\left(\tau\right)+\left(1-\lambda\right)\left(\frac{d}{d\tau}r_{2}^{f}\left(\tau\right)\right)r_{2}^{f}\left(\tau'\right)\right]\mbox{.}\ ] ] taking the expected value @xmath131 , where @xmath9 is the vacuum field and @xmath132 an arbitrary atom state , results @xmath133\right|a\right\rangle \times\\ \\
\left\langle 0\left|\left(\lambda-\frac{1}{2}\right)\left[\phi^{f}\left(\tau\right),\phi^{f}\left(\tau'\right)\right]+\frac{1}{2}\left[\phi^{f}\left(\tau\right),\phi^{f}\left(\tau'\right)\right]_{+}\right|0\right\rangle \mbox { , } \end{array}\label{eq : vf.valor.esperado}\ ] ] @xmath134 @xmath135\right|0\right\rangle \times\\ \\
\frac{d}{d\tau}\left\langle a\left|\frac{1}{2}\left[r_{2}^{f}\left(\tau\right),r_{2}^{f}\left(\tau'\right)\right]_{+}-\left(\lambda-\frac{1}{2}\right)\left[r_{2}^{f}\left(\tau\right),r_{2}^{f}\left(\tau'\right)\right]\right|a\right\rangle \mbox{. } \end{array}\label{eq : rr.valor.esperado}\ ] ] in equations and we can recognize the commonly known functions in quantum optics as correlation function @xmath136 and linear susceptibility @xmath137 .
for the field , the correlation and susceptibility functions are @xmath138_{+}\right|0\right\rangle \mbox{,}\label{eq : func.correl.campo}\ ] ] @xmath139\right|0\right\rangle \mbox{,}\label{eq : func.suscep.campo}\ ] ] and for the atomic system @xmath140_{+}\right|a\right\rangle \mbox{,}\label{eq : func.correl.atomo}\ ] ] @xmath141\right|a\right\rangle \mbox{.}\label{eq : func.susep.atomo}\ ] ] explicit calculation of the correlation and susceptibility functions gives , to the atomic system , @xmath142 @xmath143 and , using wightman functions , to the field @xmath144}+\frac{1}{\left[\left|\delta\mathbf{x}\right|^{2}-\left(\delta t+i\epsilon\right)^{2}\right]}\right]\mbox{,}\label{eq : func.correl.campo.explicita}\ ] ] @xmath145}-\frac{1}{\left[\left|\delta\mathbf{x}\right|^{2}-\left(\delta t+i\epsilon\right)^{2}\right]}\right].\label{eq : func.suscep.campo.explicita}\ ] ] to go beyond and one must choose a parametrization , thus , using the uniformly accelerated case as before results @xmath146\mbox{,}\label{eq : func.correl.campo.explicita.inercial-1}\ ] ] @xmath147\mbox{.}\label{eq : func.suscep.campo.explicita.inercia-1}\ ] ] unfortunately , in this case , the integrals found in and are divergent .
one option to overcome this situation is to restrict the study to the asymptotic case , @xmath148 , leading to @xmath149\mbox{,}\label{eq : vfi.final}\ ] ] @xmath150\mbox{,}\label{eq : rr.final}\ ] ] where @xmath151 . nevertheless , the integral @xmath152,\ ] ] is still diverging . from equations and can be concluded two important facts . for the physical observable
associated with vacuum fluctuations and radiation reaction be independent and represented by finite quantities it is necessary to perform operators ordering .
it s also clear that the only possible order to yield not divergent rates is the symmetrical one , @xmath153 .
thus , for the symmetric case , @xmath154\mbox{,}\label{eq : vf.rindleri.final.simetrico}\ ] ] @xmath155\mbox{.}\label{eq : rr.rindler.final.simetrico}\ ] ] from , we notice that the radiation reaction is responsible for giving energy to the field and has a purely dissipative nature , independent of @xmath1 and @xmath64 . to better interpret the vacuum fluctuations , we replace @xmath110 for its explicit form @xmath156\left[\left|\left.\left\langle a\right|+\right\rangle \right|^{2}-\left|\left.\left\langle a\right|-\right\rangle \right|^{2}\right]\mbox{,}\ ] ] therefore , in the case where @xmath157 , i.e. the atom is in the excited state , results @xmath158\mbox{,}\ ] ] and , as expected , the vacuum fluctuation is responsible for giving energy . in case that @xmath159 , which is equivalent to the ground state , @xmath160\mbox{,}\ ] ]
so the vacuum fluctuation is responsible for excite the atom .
note that , on average , vacuum fluctuations cause a zero energy change , as expected , since it is related to the free part of the field and so taking place regardless of the interaction .
therefore , it is not unusual to expect that the average variation caused by it to be zero . returning to equations and ,
observe that when adding both @xmath161\left[1+\frac{2}{e^{\frac{2\pi\omega_{0}}{\alpha}}-1}\right]\mbox{,}\label{eq : flutua=0000e7ao.total.final}\ ] ] and the equation has no dependence of the parameter @xmath162 , that is , for the total physical process any ordering results in the same energy variation @xcite . keeping this in mind , we now establish the following correspondences between and : @xmath163 here
, equation is projected in the basis of the eigenstates of the atomic system .
thus , taking the expectation value of equation in an arbitrary state @xmath132 of the atomic system and defined the contracted notation @xmath164 equation becomes @xmath165+h.c.\,\mbox{,}\label{eq : eq.mestra.2.niveis}\ ] ] where @xmath166 represents the hermitian conjugate .
equation is , in this case , @xmath167 . replacing the fields , we obtain @xmath168\mbox{,}\ ] ] or , using the results found for the expected value of the number operator , @xmath169^{-1}e^{i\omega_{\mathbf{\mathbf{k}}}\left(\tau'-\tau''\right)-i\mathbf{k}.\left(\mathbf{x'-x''}\right)}+\left[1-e^{-\beta\omega_{\mathbf{k}}}\right]^{-1}e^{-i\omega_{\mathbf{\mathbf{k}}}\left(\tau'-\tau''\right)+i\mathbf{k}.\left(\mathbf{x'-x''}\right)}\right]\mbox{.}\label{eq : g.tau.temperatura}\ ] ] next ,
consider the master equation in the inertial case and , subsequently , in the case of an accelerated frame . in the inertial case , @xmath170^{2}}\\ \\
= \frac{\sqrt{1-v^{2}}\left[coth\left(\frac{(v-1)}{\sqrt{1-v^{2}}}\frac{\pi}{\beta}\left(\tau'-\tau''\right)\right)+coth\left(\frac{(v+1)}{\sqrt{1-v^{2}}}\frac{\pi}{\beta}\left(\tau'-\tau''\right)\right)\right]}{8\pi\beta v\left(\tau'-\tau''\right)}\mbox{. } \end{array}\ ] ] the above expression is too complicated to be integrated .
however , in the limit @xmath171 , @xmath172^{2}}=-\frac{1}{4\beta^{2}}csch\left(\frac{\pi\left(\tau'-\tau''\right)}{\beta}\right)^{2}\mbox{.}\ ] ] in this case , equation becomes @xmath173^{-2}\left[e^{ai\omega_{0}u}\sigma_{a_{aa}}\left(\tau\right)-e^{-ai\omega_{0}u}\sigma_{a_{bb}}\left(\tau\right)\right]\mbox{.}\label{eq : eq.mestra.integral.minkoski-1}\ ] ] writing the master equation for the states @xmath174 e @xmath175 , results @xmath176\right\ } , \label{eq : delta.+}\ ] ] @xmath177\right\ } \mbox{.}\label{eq : delta.-}\ ] ] from e , it follows that @xmath178 at steady regime , @xmath179 , we find the detailed balance condition @xmath180 solving the coupled equations and , with the initial conditions@xmath181 e @xmath182and the boundary conditions @xmath183 , we find @xmath184\mbox{,}\label{eq : evolu=0000e7ao.eq.mestra.minkoswki.sigma+.cte}\ ] ] @xmath185\mbox{.}\label{eq : evolu=0000e7ao.eq.mestra.minkoswki.sigma-.cte}\ ] ] the above equations and exhibit a non - trivial intertwining among @xmath186 , @xmath187 , @xmath64 that determines the time evolution of the system .
figure [ eq.mestra.minkowski ] depicts two distinct situations .
the left curves correspond to the high temperature behaviour of the system , @xmath188 ; in this case , the system evolves from a state in which @xmath189 , reaching an equilibrium state where both levels are equally filled . at low temperatures , as illustrated by the graph at right for @xmath190 , when there are initially many excited particles , @xmath191 , the system evolves in such a way to completely fill @xmath192 .
the intermediate conditions vary between these two extremes .
also , it turns out that the @xmath187 is primarily responsible for determining the values of @xmath193 for which the system converges , the parameter @xmath64 giving the rate in which it converges . finally , in the limit @xmath194
, we obtain in both situations @xmath195 @xmath196 we then conclude that the two - level atomic system obeys the fermi - dirac statistics , as expected , according to bloch theorem .
considering now the uniformly accelerated frame with parametrization , we obtain @xmath197^{2}}\\ \\ \\
= \dfrac{\alpha\left[coth\left(\frac{\pi\left(e^{\alpha\tau'}-e^{\alpha\tau''}\right)}{\alpha\beta}\right)-coth\left(\frac{2\pi e^{-\frac{1}{2}\alpha\left(\tau'+\tau''\right)}sinh\left(\frac{1}{2}\alpha\left(\tau'-\tau''\right)\right)}{\alpha\beta}\right)\right]}{8\pi\beta\left[cosh\left(\alpha\tau'\right)-cosh\left(\alpha\tau''\right)\right]}\mbox{. } \end{array}\label{eq : eq.mestra.integral.rindler}\ ] ] as in the inertial case , the equation is too complex to be integrated .
however , limiting cases provide us remarkable results .
thus , in the limit @xmath198 , we obtains @xmath199 recovering , as expected , the inertial result and assuring the correctness of @xmath200 . in the limit of low temperatures , ie , @xmath201 ,
we find @xmath202 by comparing with , we obtain the relation proposed by unruh , @xmath203 i.e. the same result holds both for an accelerated observer at zero temperature as for an inertial observer immersed in a thermal bath at temperature @xmath40 .
though , looking at the graph in figure[g.tau ] to @xmath200 we found that the behaviour regarding the temperature and acceleration are not the same .
note that , at zero temperature , as the acceleration increases , @xmath200 grows almost linearly .
however , fixing a non - zero temperature , results that by increasing the acceleration @xmath200 decreases .
we conclude that the temperature influences the way the proper acceleration behaves .
in this paper , after introducing the original motivation behind the so - called unruh - fulling - davies - dewitt effect through the response function of a detector , which undergoes hyperbolic motion , we have shown that the thermofield dynamics ( ftd ) formalism revealed itself as a natural scenario for describing the behaviour of accelerated systems at finite temperature . in such approach , the thermal states belong to a hilbert space in the number representation and describe a microscopic system interacting with a _ reservoir _ , leading to equations of motion for both the two - level detector and the radiation field , so that in the coarse - graining approach , the dynamics of these systems consist of a markovian process . supposing that the sub - systems reach a global thermal equilibrium at a common temperature , we have shown that the local fluctuation due to admissible couplings , enlarging the class of dewitt detectors , lead to a master equation for the population levels whose dynamics respect the fluctuation - dissipation theorem . as a result
, we arrived at the conclusion that the two - point correlation function associated to the population levels depend on both the _ reservoir _ equilibrium temperature and the magnitude of the hyperbolic acceleration , indicating the existence of a range of temperatures in which the population of the excited state may also decrease at increasing acceleration magnitudes @xcite and , therefore , unruh and _ reservoir _ temperatures do not coincide , although the detailed balance condition is respected .
this is corroborated by the analysis of the vacuum fluctuation and radiation reaction hermitian contributions to the energy variation rates of the scalar field considered as a thermal _ reservoir _ by choosing a suitable operator ordering .
so , we have found that the evolution equation for an accelerated observer coincides with the corresponding one for an inertial observer immersed in a thermal unruh bath if , and only if , one ascribes _
a priori _ a zero value temperature to the radiation field in rindler coordinates .
in section [ sec : the - master - equation ] the operator @xmath204 has been replaced by @xmath205 , which is equivalent to expanding the master equation only to second order .
if applied recursively , this procedure will generate contributions to higher orders , appearing terms of the type triple , quadruple commutators , and so on .
the formal convergence analysis of this series is of extreme complexity , being out of the scope of this paper .
however , we present elements of plausibility , which indicate that , at least , the series should asymptotically converge . the exact expression for the population is @xmath206\right]\mbox{.}\ ] ] integrating the equation for @xmath207 between @xmath208 and @xmath209 , @xmath210\mbox{,}\ ] ] and substituting in the above equation naturally gives rise to the second and third order terms in the expansion , @xmath211\right]\\ \\
+ \left(\frac{1}{i\hbar}\right)^{3}\int_{t}^{t+\delta t}dt'\int_{t}^{t'}dt''\int_{t}^{t''}dt'''tr_{r}\left[v\left(t'\right),\left[v\left(t''\right),\left[v\left(t'''\right),\rho\left(t'''\right)\right]\right]\right]\mbox{. } \end{array}\ ] ] due to the shape of the potential , we find that the traces can be factored into a part referring to the system @xmath30 and the other , @xmath212 , to the system @xmath47 .
this will allow us to ignore the higher order terms .
we must show that , after a time interval @xmath213 sufficiently large , the three points correlation of the _ reservoir _ observable are more strongly suppressed .
projecting on the @xmath47 basis we have @xmath214=z_{r}^{-1}\underset{m , n}{\sum}e^{-\beta e_{n}}\left\langle n\right|r^{s}\left|m\right\rangle \left\langle m\right|r^{s}\left|n\right\rangle e^{-i\omega_{nm}\left(t'-t''\right)}\mbox{,}\ ] ] @xmath215\\ \\
= z_{r}^{-1}\underset{m , n , l}{\sum}e^{-\beta e_{n}}\left\langle n\right|r^{s}\left|m\right\rangle \left\langle m\right|r^{s}\left|l\right\rangle \left\langle l\right|r^{s}\left|n\right\rangle e^{-i\omega_{nm}\left(t'-t''\right)}e^{-i\omega_{nl}\left(t''-t'''\right)}\mbox{. } \end{array}\ ] ] we can not go further without additional information about the _ reservoir_. however , the _ reservoir _ has dense , almost continuous , spectrum of energy so we assume that the different phases in @xmath216 cause a destructive interference as time increases .
thus , @xmath216 oscillates very rapidly and its contribution to the integral becomes smaller .
considering two different time scales , such that @xmath217 where @xmath218 is the time correlation between observables of the _ reservoir _ and @xmath219 the time evolution associated with the microscopic system , we have at second order , @xmath220 close to a given dominant frequency @xmath221 in the expression , @xmath222 is the typical value of the interaction potential at equilibrium , the observable microscopic system is taken at a given time , around which it does not vary appreciably within the integration intervals . performing the integral @xmath223\ ] ] @xmath224\delta t\mbox{;}\ ] ] therefore , @xmath225 implying @xmath226 to third order we have @xmath227 resulting @xmath228 then , in weakly coupled systems , the contribution of higher order terms is less than the lower order ones , even though the series does not converge uniformly , being only asymptotic .
in this appendix we ask ourselves what could happen if instead of , we had a derivative coupling .
thus , we propose a more general interaction given by @xmath229 where the constant @xmath230 is introduced to ensure the correct dimension . with the interaction ,
we find new equations for the vacuum fluctuations and radiation reaction , equivalent to the equations and , @xmath231\mbox { , } \end{array}\label{eq : vf.derivativo}\ ] ] @xmath134 @xmath232\times\\ \\
\left[\frac{1}{2}\,sin\left(\omega_{0}\left(\tau-\tau'\right)\right)+i\left(2\lambda-1\right)cos\left(\omega_{0}\left(\tau-\tau'\right)\right)\left\langle a\left|r_{3}^{f}\left(\tau_{0}\right)\right|a\right\rangle \right]\mbox{. } \end{array}\label{eq : rr.derivativo}\ ] ] choosing the symmetrical order , @xmath153 , we eliminate the divergences both in and .
then , by solving the integrals present in and , we recover and for the non - derivative case .
we thus observe that both results are independent of the coupling order , always giving the same energy variations .
in section [ sub : master-2-level ] , we derived an equation for the energy variation rate of the atomic system due to vacuum fluctuations and radiation reaction . in this appendix , we show that the energy balance is respected , observing that the energy variation of the field is equal , in module , to that of the atomic system .
we start from the heisenberg equation @xmath233=i\mu r_{2}\left(\tau\right)\left[\phi\left(\tau\right),h_{f}\left(\tau\right)\right]\mbox{.}\ ] ] taking the expectation value in the field vacuum @xmath9 and the atomic state @xmath132 , we obtain @xmath237\right|a\right\rangle \partial_{\tau}\left\langle 0\left|\left[\phi^{f}\left(\tau\right),\phi^{f}\left(\tau'\right)\right]_{+}\right|0\right\rangle \mbox{,}\ ] ] @xmath238_{+}\right|a\right\rangle \partial_{\tau}\left\langle 0\left|\left[\phi^{f}\left(\tau\right),\phi^{f}\left(\tau'\right)\right]\right|0\right\rangle \mbox{,}\ ] ] where we can recognize the correlation functions and susceptibility , , and , | in this work we study the behaviour of two weakly coupled quantum systems , described by a separable density operator ; one of them is a single oscillator , representing a microscopic system , while the other is a set of oscillators which perform the role of a _ reservoir _ in thermal equilibrium . from the liouville - von neumann equation for the reduced density operator
, we devise the master equation that governs the evolution of the microscopic system , incorporating the effects of temperature via thermofield dynamics formalism by suitably redefining the vacuum of the macroscopic system . as applications
, we initially investigate the behaviour of a fermi oscillator in the presence of a heat bath consisting of a set of fermi oscillators and that of an atomic two - level system interacting with a scalar radiation field , considered as a _ reservoir _
, by constructing the corresponding master equation which governs the time evolution of both sub - systems at finite temperature .
finally , we calculate the energy variation rates for the atom and the field , as well as the atomic population levels , both in the inertial case and at constant proper acceleration , considering the two - level system as a prototype of an unruh detector , for admissible couplings of the radiation field . |
starting with the question of whether the excited - state lifetime of an atom can be modified , physicists have continuously explored ways to engineer the radiative properties of quantum emitters .
the first experimental studies were performed in the near field of flat interfaces @xcite , but the change of spontaneous emission is commonly associated with purcell s prediction that a cavity of quality factor @xmath2 and mode volume @xmath3 can accelerate the radiation of a dipolar transition by @xmath4 folds , where @xmath5 is the wavelength in the corresponding medium @xcite . following this recipe ,
many clever resonator schemes such as high - q open fabry - prot cavities ( fpc ) , monolithic fpcs in form of pillars , whispering gallery mode resonators and photonic crystal structures have been investigated for realizing large purcell factors @xcite .
nevertheless , sizable modifications of the spontaneous emission process remain nontrivial because no cavity design has succeeded in providing all the decisive ingredients of large @xmath2 , small @xmath3 ( ideally down to its fundamental value of the order of @xmath6 ) , a facile way of tuning the cavity resonance , and compatibility with emitters of different materials . of the various cavity geometries ,
open fpcs remain particularly attractive because they nicely lend themselves to the latter two criteria .
efforts using open fpcs have usually pursued large @xmath7 via high @xmath2s instead of low @xmath3 , but this choice is accompanied by several disadvantages .
first , the resulting narrow cavity linewidths make the cavity extremely sensitive to mechanically , thermally or stress - induced length changes .
second , broad transitions in the condensed phase can not couple to very narrow cavity resonances in an efficient manner .
this is especially a restriction for room - temperature studies , where homogeneous linewidths of quantum dots , molecules or ions can reach several nanometers .
in addition , scattering from finite - sized particles , e.g. from diamond nanocrystals , could quickly spoil the high q of the cavity .
furthermore , narrow cavity resonances do not allow simultaneous coupling to different transitions , for instance , for studies of @xmath8 transitions or of nonlinear interactions involving several light beams @xcite . in this article , we explore a new cavity design using a micromirror with a radius of curvature as small as 2.6 @xmath9 fabricated on a cantilever . by realizing a microcavity with a sub-@xmath0 volume and low @xmath2
, we obtain a sizable @xmath7 that would lead to the modification of the spontaneous emission rate by more than one order of magnitude while keeping a spectral bandwidth as broad as 1 thz .
the second novel feature of our arrangement is the compatibility of its high numerical aperture ( @xmath10 ) with the recent advances in efficient atom - photon coupling via tight focusing @xcite .
third , the open character , scanning capability and low q of our fpc provide the possibility of using it as a scanning microscope @xcite , which can be used for local field - enhanced spectroscopy @xcite and sensing @xcite of very different materials such as semiconductor quantum dots , carbon nanotubes , organic molecules , rare earth ions , and biomolecules . in particular , the low q of the cavity ensures that even broad resonances of these materials couple efficiently to the cavity modes at room temperature .
fourth , the cantilever - based feature of our experimental arrangement makes it highly interesting for an unexplored regime of optomechanical investigations @xcite as we discuss in the concluding paragraph of this article .
the inset in fig . [ setup](a ) shows the schematics of our cavity made of a flat distributed bragg reflector ( dbr ) and a micromirror fabricated by focused ion beam milling . here , we started with an n - doped silicon cantilever that contained a pedestal ( diameter 8 @xmath9 ) .
the area of the micromirror to be milled was divided into pixels of diameter 5 nm . by controlling the ion dose for each pixel
, we obtained a spherical surface profile , which was then polished in a final step .
figure [ setup](b ) displays an electron microscope image of a cantilever pedestal with four mirrors of different radii of curvature , and fig .
[ setup](c ) presents an exemplary topography cross section recorded with an afm . in this case
, the mirror has a radius of curvature of @xmath11 accompanied by a root - mean - square surface roughness below 5 nm .
an opening aperture diameter of 2.4 @xmath9 provides a concave mirror with numerical aperture @xmath12 following the standard defintion of @xmath13 , where @xmath14 is the angle subtended by half the mirror aperture and @xmath15 is the mirror focal length .
we note that fabrication of micromirrors has been previously reported for wet etching @xcite , laser ablation @xcite and focused ion beam milling @xcite , but the majority of the existing works report radii of curvature above ten microns , limiting both @xmath10 and @xmath3 . to our knowledge
, we provide the highest numerical aperture for a tunable microcavity reported to date .
the micromirrors were coated with 150 nm of gold followed by 50 nm of silicon dioxide as a protective layer , yielding a nominal reflectivity of 96% at @xmath16 nm .
taking into account the slight roughness of this mirror ( see figure [ setup](c ) ) , we expect the reflectivity to be reduced to 95% due to residual scattering @xcite .
the dbr consisted of 11 layers of @xmath17 stacks finished by a 22 nm layer of @xmath18 to place the field maximum at the mirror surface .
the resulting structure had a total thickness of 2.14 @xmath9 with its design band center at 710 nm and a band edge at 841 nm , leading to reflectivities of 99.9% at @xmath16 nm and 99.99% at 745 nm .
as displayed in fig .
[ setup](a ) , the microcavity was assembled with a piezoelectric transducer ( pzt ) stack for the axial displacement of the cantilever and a pzt scanner for the lateral positioning of the dbr .
we point out that the asymmetry of the mirror reflectivities causes a suboptimal impedance matching , which can be easily alleviated by using lower reflectivity dbr . in this work , however , we have not been concerned with coupling efficiencies . intuitively speaking ,
we want to operate the cavity close to the condition , where the rays from a strongly focused incident beam are retroreflected by the curved micromirror .
it follows that a useful figure of merit for mode matching becomes the opening arc half - angle @xmath19 subtended by the mirror aperture ( see the inset of fig .
[ setup](a ) ) .
we , thus , define @xmath20 as an effective numerical aperture , which amounts to 0.46 in our experiment .
we note in passing that according to common textbook knowledge , the stability of a plane - concave fpc becomes compromised as the cavity length becomes comparable to and larger than the mirror curvature @xcite . to perform spectroscopy on the microcavity , we focused laser beams at @xmath5=785 nm and @xmath5=745 nm through a microscope objective with @xmath21 , whereby the waist of the incident laser beam was adjusted to match @xmath22 .
we then scanned the axial position of the cantilever and monitored the reflected light on a photodiode . for various @xmath23 modes .
the indices @xmath24 signify the best fit obtained for the longitudinal mode number .
, width=8 ] the blue spectra in figs .
[ cavity - spectra](a ) and ( b ) plot the measured reflection from the microcavity as a function of the change in @xmath25 at 745 nm and 785 nm , respectively . here , we slightly defocused and misaligned the incident beam to obtain some coupling to the higher transverse modes .
the red lines mark the resonances associated with the hermite - gaussian modes predicted by the equation @xcite @xmath26~ , \label{cavityeqn1}\ ] ] for a plane - concave cavity with the same nominal geometrical parameters as in the experiment but with infinitely thin mirrors . here , the the term @xmath27 signifies the gouy phase within the paraxial approximation . the symbols in fig .
[ cavity - spectra](c ) and ( d ) summarize the dependence of @xmath25 on @xmath28 indices of the transverse @xmath23 modes for each longitudinal parameter @xmath24 .
if we now fit these data with the outcome of eq .
( [ cavityeqn1 ] ) , we obtain values for @xmath24 posted in the legends of fig .
[ cavity - spectra](c , d ) .
the good cumulative comparison between theory and experiment suggests that we have reached the longitudinal mode @xmath29 .
the resulting non - integer values of @xmath24 stem from the fact that eq .
( [ cavityeqn1 ] ) neglects the penetration depth of the field in the dbr mirror @xcite as illustrated in fig .
[ volume - cuts ] for @xmath5= 745 nm . the field penetration in the mirror and a considerable gouy phase of @xmath30 in our high - na cavity
make the analysis of the spectra presented in fig .
[ cavity - spectra ] nontrivial . to account for these
, we now examine the resonance condition and spectral properties of our cavity starting from basic derivations . at the wavelength of 745 nm .
the cavity is formed by a gold - coated micromirror and a flat distributed bragg reflector marked by the dashed lines .
the length of the cavity @xmath25 measured between the air - dbr interface and the apex of the curved mirror is shown in each case together with a cut throught the intensity distribution along the red vertical dashed line .
the red horizontal dotted lines mark the dbr interface.,width=321 ] the transmission spectrum of a fabry - prot cavity in the absence of absorption is given by the airy function @xcite @xmath31 where @xmath32 is the total round trip phase experienced by the light given by @xmath33 here , @xmath34 is the wavevector , @xmath35 and @xmath36 are the phase shifts introduced by the two cavity mirrors , @xmath25 is the physical separation between the mirrors , and @xmath37 is the gouy phase shift experienced by the mode after propagating the length @xmath25 @xcite .
the cavity resonance occurs when @xmath32 is an integer @xmath24 multiple of @xmath38 such that sin(@xmath39)=0 . at this point
, one can define the finesse @xmath40 as the ratio @xmath41 , where @xmath42 is the phase separation of two adjacent resonances , and @xmath43 is the full width at half - maximum ( fwhm ) of the phase through a given resonance . if this definition is applied to eq .
( [ eq : airy ] ) , one obtains @xmath44 where @xmath45 is the geometric mean of the two mirror reflectivities .
equation ( [ eq : airy ] ) provides an expression for the spectrum of the cavity .
however , in the laboratory one does not have an easy access to the phase of the light field .
instead , one either scans the frequency @xmath46 of the incident beam at a fixed @xmath25 or changes the latter at a fixed frequency . in most textbook treatments of fabry - prots , thin planar mirrors are considered , and the dependence of @xmath37 , @xmath47 and @xmath36 on @xmath25 and @xmath46 is neglected . in that case , since only the propagation phase @xmath48 changes , the spectra can be plotted as a function of @xmath25 or @xmath46 in a fully equivalent fashion , leading to @xmath49 .
here , @xmath50 and @xmath51 denote the separation of two adjacent resonance lengths and frequencies ( commonly known as the free spectral range ) , respectively , whereas @xmath52 and @xmath53 signify the fwhm of the corresponding resonance as the cavity length or frequency are varied .
this relation then allows a direct conversion of data obtained from cavity length spectroscopy to the properties of the frequency spectrum .
we now allow @xmath37 , @xmath47 and @xmath36 to depend on @xmath25 and @xmath46 .
it turns out that the dependence of @xmath47 and @xmath36 on @xmath25 is fairly negligible ( as can be inferred from fig .
[ volume - cuts ] ) and @xmath54 varies linearly with @xmath25 in the regime of our cavity parameters , as shown in fig .
[ phases](a ) . in this case , eq .
( [ eq : roundtripphase ] ) reads @xmath55 , where @xmath56 and @xmath57 denote the zeroeth and first order terms of the taylor expansion of @xmath54 , and @xmath58 signifies an effective frequency offset . in this case
, the finesse can still be calculated as @xmath59 , which is conveniently accessible in the experiment .
next , we have to relate @xmath60 and @xmath52 in order to extract the quality factor @xmath61 from our experimental measurements .
we , thus , differentiate eq .
[ eq : roundtripphase ] with respect to @xmath5 at the cavity resonance to obtain @xmath62~. \label{eq : lengthdispersion}\ ] ] the first term in the main paranthesis is the axial mode number @xmath24 of an ideal cavity , while the inner parenthesis contains corrections to it . in general , these corrections need not be small and can have implicit dependence on @xmath5 .
considering @xmath63 , we arrive at @xmath64 \mathcal{f}=q_{\rm eff}\cdot\mathcal{f}~ \label{eq : q - f}\ ] ] with @xcite @xmath65 to obtain the quantities in eq .
( [ eq : qeffexplicit ] ) , we measured the reflectivity spectrum of the dbr and compared the outcome with the calculations obtained from a transfer matrix method ( see fig .
[ phases](b ) ) . the corresponding amplitude and phase of the reflection coefficient
are shown in fig .
[ phases](c ) .
the phase shift upon reflection from the gold mirror is also calculated using the dielectric function from ref .
@xcite and is plotted in fig .
[ phases](d ) .
the phase properties of the mirrors at the working wavelengths are summarized in table [ table : mirrorphases ] .
the fit requires @xmath66 = 708.5 nm .
this slight discrepancy is likely due to finite tolerances in the dbr fabrication process .
( c ) amplitude and phase of the complex reflectivity of the dbr .
( d ) phase shift upon reflection from the gold mimrror.,width=88 ] @xmath67 & -0.5661@xmath68 & -0.1960@xmath68 + @xmath69 & -1.1405@xmath68 & -1.1310@xmath68 + [ table : mirrorphases ] we are now prepared to analyze the cavity length spectra . in fig . [ linewidths](a ) we plot a spectrum recorded at @xmath16 nm , where the incident beam was aligned to minimize the coupling to higher transverse modes .
the finesse and @xmath70 extracted from this spectrum are shown by the symbols in fig .
[ linewidths](b ) .
these data reveal that contrary to the common case of large cavities , @xmath40 and @xmath2 vary strongly with the mode number . to understand this behavior
, we set up a simple model based on the propagation of a gaussian beam between the two mirrors .
we calculated the position and size of the beam waist @xmath71 after each round trip , whereby we took into account the loss at the finite aperture of the curved mirror ( about 0.1% of the power per round trip for the second longitudinal mode ; i.e. @xmath24=2 ) simply as a scalar factor .
the red curve in fig . [ linewidths](b ) shows that the @xmath2 calculated from the cavity decay time is in good agreement with the experimental values .
we do notice deviations for the highest mode orders , where @xmath25 is large enough that our model is no longer valid , and diffraction at mirror edges and losses due to beam clipping become important .
the nonmonotonous behavior of @xmath2 in fig . [ linewidths](b ) is the result of the competition between finite - aperture losses on the one hand and gain in the photon lifetime for larger @xmath25 on the other .
the blue curve in fig . [ linewidths](b ) displays @xmath72 , which again compares well with the experimentally measured values determined from @xmath73 .
( red ) and finesse @xmath72 ( blue ) .
symbols : experiment ; curves : model .
, width=8 ] the mode volume @xmath3 plays a central role in the performance of a cavity as suggested by the purcell formula in the context of cavity quantum electrodynamics @xcite , the threshold formula in laser theory @xcite and its role in sensing @xcite .
this quantity is usually calculated as the integral of the electromagnetic energy density in the cavity mode normalized to the maximum of the field intensity . for fabry - prot resonators with perfect conductor boundaries ,
one can write @xmath74 , where @xmath75 is the gaussian mode waist . in the presence of a dbr mirror , the extra energy leakage in the dielectric layers
makes the calculation of the mode volume nontrivial , especially when @xmath25 becomes comparable with @xmath5 and the dbr penetration depth @xcite . in order to consider this effect and the curvature of the metallic mirror , we performed full - numerical eigenmode simulations ( comsol multiphysics ) to extract their modal properties . for computational simplicity , we neglected the thin protective layer on the micromirror , chose a simulation box size of about @xmath76 and imposed perfectly - matched layers and scattering boundary conditions .
the typical mesh sizes were close to @xmath77 near the micromirror , @xmath78 in the dbr and @xmath79 in the furthest regions .
figure [ volume - cuts ] displays cross sections of the intensity distribution for three different cavity lengths at @xmath5=745 nm . to evaluate the mode volume from eigenmodes in the presence of radiative losses , we followed the definition provided by ref .
@xcite , which allows for the treatment of the purcell factor in complex environments .
figure [ volumepurcell ] shows the calculated mode volume as a function of the mode number @xmath24 for @xmath80 nm . we find that the volume of the first accessible mode with @xmath29 is as low as @xmath81 , while it grows almost linearly with @xmath24 .
( magenta ) and purcell factor @xmath82 ( green ) at @xmath5=745nm.,width=283 ] the purcell factor @xmath83 is commonly used as a measure for the ability of an optical resonator to enhance the spontaneous emission rate into a well - defined mode
. an alternative formulation of the same physics is sometimes expressed as the cooperativity @xmath84 , where @xmath85 denotes the vacuum rabi frequency of an atom with transition dipole moment @xmath86 . here
, @xmath87 denotes the atomic free - space radiative decay rate , and @xmath88 signifies the cavity loss rate . after a simple manipulation , one sees that the central figure of merit remains the ratio @xmath89 , which can reach about 360 , corresponding to @xmath90 as shown in fig .
[ volumepurcell ] .
we remark that these calculations assumed a reflectivity of 94% for the gold mirror to reach @xmath2 values close to the experimental measurements .
while in this work we emphasize the advantages of the low-@xmath2 and low-@xmath3 cavity regime , higher @xmath2s can be easily achieved by using enhanced metal or dielectric layers on the micromirror instead of a gold coating . of course , one would have to keep in mind that the number of bragg layers and their dielectric contrasts would have to be chosen properly to be compatible with the high curvature of the micromirror .
we also point out that a particularly attractive application of our cavity design might be for studies at @xmath91 m , where the higher reflectivity of the gold mirror coating and smaller possible values of @xmath92 would yield higher @xmath93 .
having considered the basic features of our ultrasmall , broadband and tunable fpc , we now discuss experiments on coupling it to a point - like dipolar scatterer .
it is known that the introduction of a foreign object in the cavity adds to the overall optical path of the photons , leading to a red shift of the cavity resonance @xcite . in the case of a subwavelength nanoparticle ,
one arrives at the shift of the cavity resonance @xmath94 given by @xmath95}~ , \label{lineshifteqn}\ ] ] where @xmath96 is the electric field at position @xmath97 in the cavity @xcite . here , @xmath19 is the complex electric polarizability of the particle , which is closely linked to its absorption and scattering cross sections @xcite . in the quasi - static approximation @xmath98 where @xmath99 is the particle diameter , and @xmath100 and @xmath101 are the dielectric functions of the particle and its surrounding medium , respectively
. one might have the intuitive expectation that the introduction of a foreign object into the cavity would cause losses , thus lowering its @xmath2 .
although this is true in general , it has been shown that a nano - object can shift the cavity resonance without incurring a notable broadening @xcite .
indeed , recently there has been a great deal of activity to exploit the frequency shift of a high - q cavity for sensing nanoparticles such as viruses @xcite . to investigate the effect of a nanoparticle on our cavity
, we spin coated gold nanoparticles ( gnp ) of diameter 80 nm on the dbr with an inter - particle spacing larger than several micrometers . to ensure that we can address individual gnps
, we recorded afm images of the dbr mirror after spin coating and identified particles that were far enough from their neighbors . by matching the afm image of the particle distribution and its optical equivalent on our setup
, we could target individual gnps .
finally , we point out in passing that the polarizability of this gnp at the wavelength of interrogation is equivalent to a virus particle with diameter of 200 nm immersed in water . assuming that the particle is placed at the field maximum of a cavity , eq .
( [ lineshifteqn ] ) can be written as @xmath102 which expresses the ratio of a shift @xmath94 to the linewidth @xmath60 of the cavity resonance profile .
it is well - known that a high @xmath2 facilitates the detection of a small frequency shift experienced by a narrow resonance line . in this work ,
we enhance the shift by using very small @xmath3s so that it can be detected even for broad cavity resonances .
= 2 cavity resonance in thz as a function of the lateral position of a gold nanoparticle .
( b ) a cross section from ( a ) , displaying fwhm @xmath103 nm .
( c ) cavity resonance shift measured for different longitudinal modes.,width=8 ] figure [ shift](a ) shows the cavity shift as a function of the lateral position of a gnp , and fig .
[ shift](b ) displays a cross section from it .
we observe a shift as large as 400 ghz , equivalent to 40% of the cavity linewidth , over a gaussian lateral profile with @xmath104 nm .
to obtain these data , the dbr mirror holding the gnp was scanned laterally , while the cavity length was scanned at each pixel .
the red symbols in fig .
[ shift](c ) plot the maxima of cavity resonance shifts for different longitudinal modes . as expected from eq .
( [ lineshifteqn ] ) , the effect of the particle rapidly diminishes for higher @xmath24 modes and larger @xmath3s . to verify the measured data quantitatively ,
we have fitted them with eq .
( [ lineshifteqn ] ) , leaving @xmath19 as a free parameter .
the blue curve shows the best fit obtained for @xmath105 , which is 1.17 times larger than its expected value for a nominal gnp with a diameter of @xmath106 nm ( as specified by the manufacturer ) , @xmath107 of gold obtained from ref .
@xcite , and @xmath108 .
however , it should be born in mind that the exact knowledge of @xmath19 for a given gnp is highly nontrivial .
first , near - field coupling to the dbr surface modifies the gnp plasmon resonance and polarizability @xcite . by using an analytical expression @xcite ,
we have estimated the polarizability of the gnp to increase by 1.1 times in the presence of the dbr upper layer alone .
furthermore , variations in shape , the finite size of the particle and the resulting radiation damping effect @xcite enter beyond the simple expression of eq .
( [ alpha ] ) . considering these effects ,
our experimental findings are in excellent agreement with theoretical expectations .
next , we turn to the effect of the nanoparticle on the cavity finesse for various mode orders .
since the nanoparticle scatters some of the light out of the cavity mode and has a finite absorption cross section , one can expect a degradation of @xmath72 .
figure [ q](a ) displays a map of @xmath72 as a function of the gnp lateral position in the @xmath29 mode , and the top curve in fig .
[ q](b ) plots a cross section from it .
we find that @xmath72 decreases by about 7% from 60 to 56 in the presence of the particle while this behavior is substantially changed for higher modes ( see the other plots in fig .
[ q](b ) ) . in fact , we find that the particle can even improve @xmath72 .
for example , it is increased by about 10% from 6.5 to 7 for @xmath109 . as a function of the particle position for @xmath29 .
( b ) cross sections of @xmath72 vs gnp position for different cavity modes . the mode number @xmath24 is indicated in each measurement .
( c ) a two - dimensional map of the cavity coupling , obtained by normalizing the reflected power to the incident power , as a function of the particle position .
( d ) cross sections of the cavity coupling efficiency for different @xmath24s.,width=321 ] for the shortest cavities , the mirrors are so close that the light is efficiently captured after each round trip . in this case ,
absorption by the particle determines the quality factor and finesse .
as @xmath25 approaches @xmath110 , the unperturbed resonator becomes less stable and @xmath72 is lowered ( see fig .
[ linewidths](b ) ) .
interestingly , in this regime the addition of a gnp ameliorates the situation . to investigate the modification of the finesse by the nanoparticle further
, we have performed numerical calculations that include the nanoparticle . here
, we consider the contribution of losses caused by absorption and scattering ( signified by @xmath111 and @xmath112 ) to the cavity quality factor according to @xmath113 .
we , thus , first estimate the absorption and scattering power from simulations by calculating the overall resistive losses per optical cycle and the power flow in the directions perpendicular to the cavity axis .
figure [ scatteringcoefficients_finesse](a ) shows these quantities for the different modes normalized to the cavity energy per optical cycle .
we find that without the nanoparticle , absorption losses decrease while the scattering losses grow with with @xmath24 .
a nanoparticle on the cavity axis and in contact with the dbr slightly increases the absorption ( compare red and black lines ) , but it reduces the power scattered outside the cavity ( green line ) in comparison with the case without it ( blue ) .
these results confirm that the gnp acts as a mode matching antenna to improve the coupling into the cavity mode .
to support this hypothesis further , we measured the amount of light that circulates inside the cavity by monitoring the power reflected from the dbr .
figure [ q](c ) plots the two - dimensional map of the coupling efficiency for @xmath29 , and fig .
[ q](d ) displays the cross sections for various modes .
the similarity of the patterns of the two data columns in fig .
[ q](b , d ) shows that the gnp strongly influences the coupling of the incoming laser beam into the cavity mode . at this point , we note that one can also perform interferometric scattering measurements ( iscat ) to detect individual gnps without the micromirror @xcite . using this method
, we determined a fwhm of 690 nm for the focus spot of the incident beam in the absence of the cantilever .
this value agrees to within 10% with the fwhm that we found from fig .
[ shift](b ) , indicating that the cavity mode waist is indeed well matched to the incoming spot size .
the black symbols in fig .
[ scatteringcoefficients_finesse](b ) display the simulated cavity finesse , whereby we adjusted the reflectivity of the gold mirror to @xmath114 to reduce the finesse by 2.4 times from its expected value for the ideal structure in order to emulate the experimental values .
the trend obtained from the numerical simulations agrees very well with that presented in fig .
[ linewidths](b ) .
in particular , we observe that the cross - over between the absorption and scattering losses at @xmath24=3 ( see fig .
[ scatteringcoefficients_finesse](a ) ) results in larger finesse in the presence of the nanoparticle ( green ) for higher modes , verifying the experimental observations in fig .
in fact , even the magnitude of the change in @xmath40 agrees well with the experimental findings .
nm nanoparticle ( green symbols ) and without it ( black).,width=321 ]
the cavity regime studied in this work has brought forth different interesting phenomena , many of which had not been encountered in previous works on microcavities .
these include a strong dependence of the cold cavity @xmath2 and @xmath72 on the longitudinal mode order @xmath24 , the largest numerical aperture reported to date and a wavelength - sized mode waist , increase of cavity @xmath72 by a nanoparticle , the combination of a sub-@xmath0 mode volume and frequency tunability , and the fact that one mirror sits on a cantilever .
these features are very promising for a number of future studies ranging from biophysics to quantum optics .
one of our future goals is to modify the branching ratio of solid - state emitters such as organic molecules or color centers .
the energy level scheme of such systems usually involves a narrow transition on the so - called zero - phonon line ( zpl ) accompanied by a broad phonon wing and transitions among other vibrational levels . in order to have a strong transition for quantum optics experiments ,
it is desirable to exploit the purcell effect to enhance the zpl and thus improve its branching ratio with respect to the other transitions . in our current measurements , @xmath115 reaches a value of @xmath116 , corresponding to @xmath117 , which would be sufficient for turning a typical branching ratio of about 30% for the zpl of aromatic molecules to about 92% .
such a nearly perfect two - level system can then act as an efficient source of single narrow - band photons @xcite .
another potential use of our cavity concept is for the achievement of few - photon nonlinear effects on single quantum emitters @xcite .
examples of such an effect are three - photon mixing , stimulated rayleigh scattering and hyper - raman scattering @xcite . while there have been a few reports of direct cavity - free optical nonlinear studies on single molecules and quantum dots @xcite , enhancements of the optical field in the cavity and of the radiation of a quantum emitter into its mode can boost these effects further .
here , a large cavity bandwidth is of crucial importance since simultaneous coupling to several wavelengths would be otherwise not possible .
the broad resonance of a microcavity is also a great asset for studying cooperative effect among many quantum emitters with slightly different resonance frequencies or with large homogeneous lines as it occurs in solid - state systems .
an example of such phenomena concerns rabi splittings and coupling to polaritonic states at room temperature @xcite .
although the coupling of each individual molecule or exciton is negligible in these systems , strong rabi splitting are observed when a large number of emitters join efforts .
the combination of a large cooperativity , small @xmath3 and low @xmath2 would allow one to examine polaritonic physics with far fewer emitters than previously accessible both at room temperature and at cryogenic conditions .
a further practical advantage of a considerable cooperativity factor combined with a low @xmath2 is that the experimental setup becomes much less sensitive to mechanical instabilities .
this is particularly helpful for experiments in high vacuum or in cryostats , where vibrations are omnipresent .
indeed , the arrangement of the cavity presented here could be highly interesting for use in atom @xcite and nanoparticle trapping @xcite , especially for on - chip schemes @xcite .
for example , the antinode in fig .
[ volume - cuts](a ) is ideally suited for this purpose , keeping the particle sufficiently far from mirror surfaces to avoid van der waals forces @xcite .
the cantilever - based nature of our experimental arrangement also holds great promise in the context of optomechanics .
let us consider a silicon cantilever of thickness 0.7 @xmath9 , length 10 @xmath9 and width 5 @xmath9 , yielding a mechanical oscillation frequency of about 5 mhz and mass of @xmath118 kg .
a micromirror on such a cantilever forming a cavity length @xmath119 would result in a nearly maximal field per photon @xmath120 and an optomechanical coupling strength between the photon and a single phonon of @xmath121 hz , which is comparable with the best reported values @xcite .
for these experiments , a higher q would further improve the cavity performance . as mentioned earlier ,
these can be readily increased by using dielectric coatings instead of gold . to respect the small radius of curvature of the micromirror
, one would use materials with strong dielectric contrasts to reduce the number of the necessary bragg layers .
other applications of the work presented here are in optical sensing and analytics . over the past decades ,
a large body of literature has been devoted to the use of optical micro- and nanostructures such as waveguides , cavities and plasmonic nanoantennas for detecting small traces of biological matter @xcite .
the existing works on microcavity sensing use fairly large mode volumes and require , therefore , high @xmath2s to reach sufficient sensitivity .
recently , we demonstrated that single small proteins can be detected via their direct scattering in a label - free fashion if only diffraction - limited imaging is implemented @xcite . in a nutshell
, we have shown that the scattering cross section of a protein , as small as it is , is large enough to extinguish a measurable amount of power from a laser beam .
if one now maintains a tight focus and circulates the laser beam in a cavity to allow for repeated interactions with the analyte species , one directly wins in the detection sensitivity by @xmath122 folds . based on a similar argument
, our tunable and scannable microcavity can also find applications in raman microscopy , where small cross sections would be compensated by larger excitation intensities .
again , a large cavity bandwidth is essential to allow a simultaneous coupling of the excitation beam and the raman signal .
the physics of microlasers is another area that would benefit from our cavity design . since its invention more than half a century ago
, there has been a continuous stream of reports on various fundamental aspects of laser physics , for example of lasing with minimal gain or with very small thresholds .
it turns out that again an important figure of merit is given by @xmath93 @xcite .
here , a higher @xmath2 increases the degree of coherence although it also minimizes the overlap between the homogeneous spectrum of the gain medium and the cavity mode , resulting in higher thresholds .
a lower @xmath3 ( and thus a higher finesse ) pushes away the modes to reduce mode competition .
in addition , a large purcell factor ensures a high @xmath123 , which determines the fraction of the overall spontaneous emission funneled in the mode of interest according to @xmath124 .
therefore , the regime of our microcavity , simultaneously keeping a low @xmath2 , low @xmath3 and high @xmath7 , has promise in the development of new microlasers , especially in combination with fluidics @xcite .
aside from their academic interest , low - threshold microlasers are also highly in demand for applications , where miniaturization and scaling play a role .
we acknowledge financial support from the european union ( erc advanced grant singleion and project univsem of the seventh framework program under grant agreement no .
280566 ) , alexander von humboldt foundation ( humboldt professur ) and the max planck society .
we thank harald haakh , ioannis chremmos and florian marquardt for fruitful discussions and maksim medvedev , lothar meier and marek piliarik for technical support at various stages of this project .
we also thank sara mouradian for her contribution to the very early stage of this work . h.k . and
d.w . contributed equally to this work . | we report on the realization of an open plane - concave fabry - prot resonator with a mode volume below @xmath0 at optical frequencies .
we discuss some of the less common features of this new microcavity regime and show that the ultrasmall mode volume allows us to detect cavity resonance shifts induced by single nanoparticles even at quality factors as low as @xmath1 . being based on low - reflectivity micromirrors fabricated on a silicon cantilever , our experimental arrangement provides broadband operation , tunability of the cavity resonance , lateral scanning and promise for optomechanical studies . |
Subsets and Splits
Data Mining Abstracts
This query searches for and retrieves 2000 documents containing the phrase 'data mining' in the abstract, providing limited but specific document samples.